{
  "audited_at": "2026-10-05T07:00:44.196493+00:00",
  "scope": "Installed AeroSim Delivery SKILL.md and all nine internal Markdown references compared with declared external skill entry points; selected procedural references inspected separately. Remote identity comparison is not evidence of semantic equivalence.",
  "declared_external_skills": [
    "corp-v1-main",
    "corp-v1-channel-general-aerosim",
    "corp-v1-channel-releases-aerosim",
    "home-v1-discord",
    "documentation-docusaurus-aerosim",
    "corp-v1-glossary",
    "self-hosted-service-account-recovery",
    "home-v1--truenas",
    "development-durable-wave-execution",
    "development-gates",
    "development-integrity",
    "development-method-ttd",
    "development-branching-strategy-aerosim",
    "development-monorepo-pnpm",
    "development-scripts-aerosim",
    "devsecops-ci-cd-gitea",
    "devsecops-ci-cd-gitea-aerosim",
    "development-gitops-argo-cd-gondor-v1"
  ],
  "sources": [
    {
      "skill": "corp-v1-channel-delivery-aerosim",
      "repository": "corp-v1-aerosim-skills-code-agent/corp-v1-channel-delivery-aerosim",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "d8837caac7eac8438bff0d48511031168cfee5f5d631f07bffdbf34626f1cc86",
          "lines": 190,
          "remote_sha256": "71248a1d4dd733de524e0dcec55147eec16ff45591f7554b80033effe614fd38",
          "matches_remote": false,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/corp-v1-channel-delivery-aerosim/src/commit/0c0e68bde3189fd96415af6035200cf7fcb9c877/SKILL.md",
          "excerpts": [
            {
              "line": 13,
              "text": "# AeroSim Delivery Channel"
            },
            {
              "line": 15,
              "text": "<!-- AEROSIM_BLOCKER_OWNERSHIP:BEGIN -->"
            },
            {
              "line": 16,
              "text": "## Hermes owns blocker resolution"
            },
            {
              "line": 18,
              "text": "RootAtSkic's [project-wide correction](https://discord.com/channels/1518726359512387766/1537227683191197807/1554922326859063327) makes Hermes main session accountable for resolving blockers and coordinating workers, not merely reporting blocked cards or asking the human to manage the queue. Preserve this project-local rule during adoption and synchronization."
            },
            {
              "line": 20,
              "text": "Apply the **Hermes owns blocker resolution** procedure in `corp-v1-channel-general-aerosim` (load it before handling blockers). Diagnose and repair within existing authority, check standing permissions before asking again, reconcile approved bookkeeping through the owning workflow, and follow cross-channel handoffs through to resolution. Reuse existing cards/workers; do not generate duplicate graphs or report queued work as running. Escalate only the exact decision, unavailable capability or authority boundary Hermes cannot resolve. Preserve human domain ownership and all security, review, UAT and release gates. Verify resumed execution rather than treating a requeue as a fix."
            },
            {
              "line": 21,
              "text": "<!-- AEROSIM_BLOCKER_OWNERSHIP:END -->"
            },
            {
              "line": 24,
              "text": "## Standing AeroSim PR merge authority"
            },
            {
              "line": 26,
              "text": "RootAtSkic authorizes Hermes to merge PRs across all AeroSim repositories and channels without per-PR human authorization. Source: [explicit standing delegation](https://discord.com/channels/1518726359512387766/1537544173366943886/1554873863165182015)."
            },
            {
              "line": 28,
              "text": "This project-local rule supersedes older instructions in this skill, linked references, or inherited guidance requiring a human to authorize PR merging or forbidding Hermes from merging. Preserve it during skill adoption and synchronization. Complete required review and exact-head CI, freshly verify the unchanged mergeable head, merge it, and verify the integration revision and required post-merge checks. Do not block a reviewed, green PR for owner merge approval or request repeated permission."
            },
            {
              "line": 30,
              "text": "This delegates PR integration, not product/design/architecture decisions, human UAT acceptance, or independently gated release/deployment actions. If merging a GitOps or release PR itself triggers a separately gated deployment, establish that deployment authority first; do not relabel it as a PR-merge approval. Repository protections, security checks, and genuine technical blockers still apply."
            },
            {
              "line": 34,
              "text": "## Overview"
            },
            {
              "line": 36,
              "text": "## Hermes Kanban direct-link and channel-boundary contract"
            },
            {
              "line": 38,
              "text": "The Hermes board at slug `aerosim-uat-gitops` is named **AeroSim Delivery**. It is dedicated exclusively to tasks originating from Discord channel `corp-v1-aerosim-delivery` (`1537227683191197807`). Create new cards on this board only for work requested, admitted, or recovered from this Delivery channel; do not place UAT, GitOps, UI/UX, Architecture, Releases, General, Scope, Kanban-channel, or another project's independently owned work on it. Cross-channel dependencies may be linked as external evidence but must remain in their owning channel/process rather than becoming cards here. Historical cards may remain as evidence, but they do not broaden the board's future intake boundary."
            },
            {
              "line": 40,
              "text": "Whenever a response mentions this project's Hermes Kanban board, initiative, card, or task, it must include the direct board link: [AeroSim Delivery](https://jarvis-jr-v1.lego-cloud.eu/kanban?board=aerosim-uat-gitops). Never report only the board name, slug, card count, or wording such as \u201cvisible cards\u201d or \u201cavailable in the dashboard.\u201d If no task-specific deep link exists, link the board and include the exact task ID or title in the same statement."
            },
            {
              "line": 43,
              "text": "## Dedicated Channel Workspace"
            },
            {
              "line": 51,
              "text": "Keep new operational files for this channel inside this root rather than placing them directly in `/opt/data`. Use `workspace/<TASK_ID>/` for Task worktrees and evidence. Put reusable Delivery dependencies\u2014such as pinned Node, pnpm, Nginx, and package caches\u2014in the channel-level `cache/` so later Tasks reuse them. Browser, Playwright, database, and other non-unit test dependencies belong to the UAT cache. Task-local `cache/` is only for data that is genuinely specific to that Task and must not be shared. Keep durable project records in the approved AeroSim repositories. Do not place another project's work in this folder. The folder boundary organizes local execution only and does not broaden this channel's authority."
            },
            {
              "line": 54,
              "text": "## Task Statuses"
            },
            {
              "line": 56,
              "text": "Use the exact status found in the Task file. Tasks move through `IN_BACKLOG \u2192 READY_FOR_DELIVERY \u2192 IN_PROGRESS \u2192 TO_BE_RELEASED \u2192 DONE`. `BLOCKED` is a separate flag with a reason; it does not replace the Task status. A merge or green CI does not mean `DONE`."
            },
            {
              "line": 58,
              "text": "## Selectable Option Controls"
            },
            {
              "line": 60,
              "text": "When RootAtSkic must select an option through `clarify` or another button control, explain the decision and every option in a normal message before sending the control. Label the explanations `Option A`, `Option B`, and so on; include each option's action, effects, trade-offs, and identify the recommendation there."
            },
            {
              "line": 62,
              "text": "The selectable `choices` must contain only `Option A`, `Option B`, and so on, in the same order as the preceding explanations. Keep the control question to the decision prompt; do not repeat option details in it. Never place the option explanation or effect inside a selectable button. Map each short button label unambiguously to its preceding explanation. Put the recommended option first so the control's automatic recommendation matches the prose."
            },
            {
              "line": 64,
              "text": "## Project Ownership Dependency"
            },
            {
              "line": 66,
              "text": "Before deciding ownership, approval routing, escalation, merge authority, deployment authority, or release authority, load `corp-v1-channel-general-aerosim` with `skill_view` and follow its **Project-local Human Ownership \u2014 Preserve During Adoption** section. Delivery does not duplicate that project-level ownership record."
            },
            {
              "line": 68,
              "text": "This skill adds only Delivery-specific limits: the Hermes main session owns Task selection, worker dispatch, final evidence verification, and channel communication; cron never owns a Task. Delivery may use exactly one isolated leaf subagent to execute one Task end to end under [`references/single-subagent-task-delivery.md`](references/single-subagent-task-delivery.md). The main session remains accountable for independently reading back repository, PR, CI, review, canonical flow, cache, and process evidence before reporting a gate or status. These limits do not replace or broaden the human authority defined by the General skill."
            },
            {
              "line": 70,
              "text": "This project adoption also grants Hermes main session standing authority to reactivate only its own Gitea account, `jarvis-at-skic`, when account disablement blocks authorized AeroSim Delivery work. This is a narrow self-recovery exception\u2014not authority over any other account or security setting\u2014and requires the exact procedure and verification in [`references/runtime-and-ci.md`](references/runtime-and-ci.md). No additional per-incident approval is required when every self-recovery precondition passes."
            },
            {
              "line": 74,
              "text": "This skill owns delivery health for AeroSim. It monitors the project's `general`, `scope`, `architecture`, `ui-ux`, `kanban`, `delivery`, `uat`, and `releases` channels and starts implementation only for tasks whose parent Feature has an exact `Approved for Implementation` decision and whose exact Task status is `READY_FOR_DELIVERY`. That status is the Focus-admitted pickup handoff; no separate post-status admission gate exists. Delivery claims the Task by moving it directly to `IN_PROGRESS`, then continues the claimed work in `IN_PROGRESS`."
            },
            {
              "line": 76,
              "text": "Its scope includes development, coding, unit testing, builds, continuous integration, immutable image publication, review readiness, and continuous delivery **only to the GitOps-managed development environment**. Delivery updates development desired state only in `corp-v1-aerosim/gondor-v1-aerosim-dev`, rendered from `corp-v1-aerosim/gondor-v1-tmpl-aerosim`, and verifies Argo CD synchronization and runtime health. It does not deploy by direct `kubectl`/Helm commands, modify production desired state in `corp-v1-aerosim/gondor-v1-aerosim`, or promote a release. After verified development deployment, Delivery sends the exact application commit, image digests, rendered GitOps commit, project-root and runtime Argo CD revisions/status, endpoint, and Task identity to `corp-v1-aerosim-uat`; failed UAT returns correction work to Kanban/Delivery."
            },
            {
              "line": 78,
              "text": "Delivery owns application **unit testing only** plus static/build-time validation of the application artifacts it publishes: Dockerfiles, container image contracts, publication records/workflows, and Helm chart linting/rendering/packaging. These artifact validations are Delivery CI safeguards, not UAT product suites. `corp-v1-aerosim/corp-v1-aerosim-e2e` under the UAT channel owns every functional non-unit suite and its fixtures, configurations, runners, and evidence: integration, component-integration, browser, E2E, acceptance, security-boundary, performance, runtime-budget, smoke, regression, and deployed-environment tests. Delivery must not add or run functional non-unit suites in the application repository, local Delivery workspace, or application branch/PR workflows. E2E must not build or publish application images, install Helm, render charts, or validate application artifact publication. It may consume a user or UAT failure as defect evidence, perform source-code root-cause analysis, implement the application fix with unit-first TTD, deploy the corrected development tuple, and provide that tuple for verification. RootAtSkic performs immediate human verification when requested; governed automated non-unit verification is added and executed later only in the UAT-owned E2E repository before release. Argo CD, pod-image, endpoint, and health readback prove deployment identity and operational health\u2014not that user-visible behavior is correct."
            },
            {
              "line": 80,
              "text": "Load the project-adopted `documentation-docusaurus-aerosim` skill from `https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/documentation-docusaurus-aerosim` before changing documentation structure, navigation, Markdown/MDX, Docusaurus configuration, data-backed React components, or builds."
            },
            {
              "line": 82,
              "text": "Load the global `corp-v1-glossary` skill from `https://gitea.lego-cloud.eu/home-v1-skills-code-agent/corp-v1-glossary` whenever project documentation needs to define or explain a reusable term. Maintain one definition in the project's final top-level **Glossary** area and link to it from the owning domain page; do not duplicate glossary-style explanations across channel documentation."
            },
            {
              "line": 84,
              "text": "## When to Use"
            },
            {
              "line": 89,
              "text": "- running its recurring synchronization job;"
            },
            {
              "line": 91,
              "text": "- adding or repairing unit tests only; non-unit coverage is handed to `corp-v1-aerosim-uat` and maintained in `corp-v1-aerosim-e2e`;"
            },
            {
              "line": 98,
              "text": "## Development Skill Routing"
            },
            {
              "line": 100,
              "text": "Load `development-durable-wave-execution` for wave recovery, `development-gates` for immutable review/CI/merge identity, `development-integrity` for delivery status, and `development-method-ttd` for behavior changes. For application Actions, load `devsecops-ci-cd-gitea-aerosim` for the exact AeroSim contract and `devsecops-ci-cd-gitea` for reusable Gitea/Harbor mechanics. Load the AeroSim variants `development-branching-strategy-aerosim` and `development-scripts-aerosim` when their repository structures match the Task; load the global `development-monorepo-pnpm` skill when the Task touches the matching pnpm/Turborepo layout. Use `development-gitops-argo-cd-gondor-v1` for GitOps mechanics until `development-gitops-argo-cd-aerosim` is corrected. Global skill sources are under `https://gitea.lego-cloud.eu/home-v1-skills-code-agent/<skill>`; remaining AeroSim overlays are under `https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/<skill>`."
            },
            {
              "line": 102,
              "text": "AeroSim overlays win on exact branch naming, visibility, worker allocation, review ownership, merge authority, deployment authority, and release authority. The current one-worker-per-eligible-Task model and Hermes main-session review remain authoritative; inherit only non-conflicting reusable mechanics. TTD is mandatory for behavior changes unless RootAtSkic explicitly approves a documented exception. Apply these GitOps substitutions: template `corp-v1-aerosim/gondor-v1-tmpl-aerosim`, Delivery desired state `corp-v1-aerosim/gondor-v1-aerosim-dev`, and Releases-owned production state `corp-v1-aerosim/gondor-v1-aerosim`."
            },
            {
              "line": 104,
              "text": "Generic unit-testing, script-layout, pnpm, Git, exact-head, and Argo CD procedures belong in the referenced development skills. Non-unit test implementation and execution belong to the UAT channel and `corp-v1-aerosim-e2e`. AeroSim local references retain only project authority, lifecycle, workspace, UAT handoff, and evidence overlays."
            },
            {
              "line": 106,
              "text": "## Detailed Procedures"
            },
            {
              "line": 108,
              "text": "Load the matching reference with `skill_view` before the action starts; seeing this list or remembering an earlier load is not enough:"
            },
            {
              "line": 110,
              "text": "- before selecting, starting, continuing, reviewing, merging, or handing off a Task, load [`references/task-delivery-gates.md`](references/task-delivery-gates.md);"
            },
            {
              "line": 111,
              "text": "- before reporting Task, wave, CI, blocker, or completion status, load [`references/status-and-reporting.md`](references/status-and-reporting.md);"
            },
            {
              "line": 112,
              "text": "- before claiming a Task, creating or resuming a branch, or opening a PR, load [`references/repository-entry-and-visibility.md`](references/repository-entry-and-visibility.md);"
            },
            {
              "line": 113,
              "text": "- before login, a local/background command, CI change, image publication, or Gondor work, load [`references/runtime-and-ci.md`](references/runtime-and-ci.md);"
            },
            {
              "line": 114,
              "text": "- before designing or changing application Gitea Actions triggers, branch/PR validation, image publication, or OCI Helm chart publication, load [`references/gitea-actions-policy.md`](references/gitea-actions-policy.md);"
            },
            {
              "line": 115,
              "text": "- before repairing a failure or changing Delivery-owned documentation, load [`references/failure-and-documentation.md`](references/failure-and-documentation.md);"
            },
            {
              "line": 116,
              "text": "- before dispatching or recovering an end-to-end Task worker, load [`references/single-subagent-task-delivery.md`](references/single-subagent-task-delivery.md); and"
            },
            {
              "line": 117,
              "text": "- when a gate fails, work resumes after interruption, or before reporting a blocker, merge, or completion, load [`references/common-pitfalls.md`](references/common-pitfalls.md)."
            },
            {
              "line": 119,
              "text": "Do not rely on memory or on having read a reference during an earlier Task. Load it again at the stated decision point. In main-runtime delivery, the Hermes main session reviews the frozen Task identity, branch, PR, exact revision, applicable reference rules, and complete candidate diff itself. In single-subagent delivery, the worker records the same exact-head verdict and the main session independently verifies that verdict, all external side effects, and the merged canonical state before reporting completion."
            },
            {
              "line": 121,
              "text": "## End-to-End Task Delivery Gates"
            },
            {
              "line": 123,
              "text": "Before selecting or continuing an AeroSim Task, load and follow [`references/task-delivery-gates.md`](references/task-delivery-gates.md). It defines all 6 ordered gates, their stable `GATE_###_NAME` codes, the repository and branch used at each gate, the allowed runtime, the required work, and the exit gate."
            },
            {
              "line": 125,
              "text": "Each `corp-v1-aerosim-delivery/workspace/<TASK_ID>/` starts with the required repository clones. Add Task-specific `review/`, `cache/`, `tmp/`, `logs/`, or `artifacts/` only when that Task needs them. Put reusable installed toolchains and technical dependencies in `corp-v1-aerosim-delivery/cache/`, never in one Task's directory. Shared cache may contain pinned Node, pnpm/package stores, Nginx binaries, and similar Delivery build dependencies; browser, Playwright, PostgreSQL, and other non-unit test dependencies belong to UAT; it must never contain repository worktrees, Task evidence, database data directories, credentials, or mutable project records. Reserve `corp-v1-aerosim-delivery/uploads/` for channel-provided input files. Load [`references/runtime-and-ci.md`](references/runtime-and-ci.md) for the complete layout and isolation rules."
            },
            {
              "line": 127,
              "text": "Keep one worker per Task. RootAtSkic grants standing authorization to drain the current release Focus: whenever an exact current-wave Task is `READY_FOR_DELIVERY` with satisfied dependencies, claim and deliver it without waiting for another pickup instruction. Recover an `IN_PROGRESS` Task that has no live executor instead of leaving it idle. Run independently eligible Tasks concurrently, one worker and workspace per Task, and replenish freed capacity until no eligible or recoverable Focus work remains. Each Task keeps independent branches, PRs, reviews, CI evidence, and lifecycle events. Never dispatch a Task whose dependency is still active in that wave, never let one worker own multiple Tasks, and never use main-runtime implementation concurrently with that Task's worker. Documentation claim and handoff merges are serialized. The main session owns selection, dispatch inventory, dependency control, independent external-state verification, and communication under [`references/single-subagent-task-delivery.md`](references/single-subagent-task-delivery.md). When several blocked Tasks require RootAtSkic input, discuss and resolve them one Task at a time without pausing independently executable workers. Cron never owns a Task. Delivery records `TO_BE_RELEASED` only after implementation, CI, merge, immutable image publication, **and verified deployment to the GitOps-managed development environment** for every Task that changes application code, runtime behavior, images, charts, runtime configuration, or deployable assets. Development deployment is the mandatory final Delivery gate for releasable application work; a Task instruction must not make it optional with wording such as \u201cif deployment is required.\u201d Delivery must update `corp-v1-aerosim/gondor-v1-aerosim-dev` through its reviewed GitOps workflow, verify Argo CD synchronization and health, prove pod image IDs and runtime source/chart annotations match the exact integrated artifacts, and perform bounded endpoint/readback checks before handoff. Only documentation-only, planning-only, test-migration, or internal tooling Tasks with no deployable runtime effect may use a recorded `not_applicable` development-deployment result, with the reason verified in the Task evidence. Delivery then hands the frozen development tuple to `corp-v1-aerosim-uat`; the Task remains `TO_BE_RELEASED` while UAT runs. A failed verdict returns correction work to Kanban/Delivery. Release execution and `DONE` belong to `corp-v1-channel-releases-aerosim` after required UAT evidence exists."
            },
            {
              "line": 129,
              "text": "## Hourly Worker Reconciliation"
            },
            {
              "line": 131,
              "text": "Maintain one project-level reconciliation job at a 60-minute cadence for the active release Focus. The job is a dispatcher and recovery guard, never the owner or implementer of a Task."
            },
            {
              "line": 135,
              "text": "1. Fetch and read the canonical documentation `test` branch; enumerate exact current-wave Tasks in `READY_FOR_DELIVERY` or `IN_PROGRESS` and validate dependencies."
            },
            {
              "line": 136,
              "text": "2. Reconcile each eligible Task against external execution evidence: task-scoped worker marker, live task-owned process, workspace branch and dirty state, remote PR/head, exact-head CI, and repository or PR activity during the preceding 60 minutes. Canonical `IN_PROGRESS` alone is not evidence of a live worker."
            },
            {
              "line": 137,
              "text": "3. Keep exactly one worker per eligible Task. Do not dispatch when a non-stale worker marker or verified live task-owned process exists. A marker becomes suspect after 120 minutes without forward repository, PR, CI, or canonical-flow evidence; inspect before replacing it."
            },
            {
              "line": 138,
              "text": "4. If a `READY_FOR_DELIVERY` Task has satisfied dependencies and no worker, claim it and dispatch one isolated leaf worker. If an `IN_PROGRESS` Task has no live executor or forward evidence, recover it from its durable workspace, branch, PR, CI state, and canonical checkpoint with one replacement leaf worker."
            },
            {
              "line": 139,
              "text": "5. Write or refresh task-scoped worker markers under `/opt/data/corp-v1-aerosim-delivery/workspace/.worker-reconciler/`; include Task ID, worker/run identity, dispatch time, observed branch/PR/head, and state. Remove the marker only after verified `TO_BE_RELEASED`, verified `DONE`, explicit blocking, or supersession by a recorded replacement."
            },
            {
              "line": 142,
              "text": "8. Report only changes, recovered/stuck work, blockers, or dispatch actions. A healthy no-change run must be concise and must not mutate lifecycle state."
            },
            {
              "line": 144,
              "text": "The scheduler must not merge, release, mark `DONE`, weaken gates, or start a duplicate worker. The Hermes main session remains accountable for exact-head review, external side-effect readback, completion evidence, and channel communication."
            },
            {
              "line": 146,
              "text": "After `GATE_005_TASK_POST_MERGE` passes and the canonical Task says `TO_BE_RELEASED`, Delivery **must post a separate release-handoff message inside Discord channel `corp-v1-aerosim-releases` (`1537388451501056081`)** stating that release work can continue there. A channel mention in the Delivery channel is not sufficient. Read the new Releases-channel message back by ID and retain its channel ID and message ID as completion evidence; do not report Delivery complete until that verified post exists. The handoff message must identify the Task, its exact `TO_BE_RELEASED` state, the passed gate, and linked application/documentation PR evidence. Posting the handoff routes the team to Releases but does not authorize release execution, deployment, or the `DONE` transition."
            },
            {
              "line": 148,
              "text": "## Proactive Iteration Requirement"
            },
            {
              "line": 150,
              "text": "Every delivery iteration must complete at least one useful, safe activity for an implementation-approved Feature or delivery-health need: implement an unblocked task, add unit tests, diagnose/fix CI, improve build tooling, update real Ways of Working/Engineering guidance, verify dependency readiness, or document a concrete blocker with evidence. Never create filler changes. Treat current-release Focus as a standing work queue: immediately select exact `READY_FOR_DELIVERY` Tasks with satisfied dependencies, claim them as `IN_PROGRESS`, and continue through `TO_BE_RELEASED` without waiting for a new user prompt. If an `IN_PROGRESS` Task has no live executor, recover it from its durable workspace, branch, PR, and canonical checkpoint. Keep independently eligible Tasks progressing concurrently, one worker per Task. Human-help conversations remain sequential: present one blocked Task, resolve its decision or input, then present the next blocker. Keep architecture's approved task/dependency plan and project documentation current."
            },
            {
              "line": 152,
              "text": "## Approved Information Boundary"
            },
            {
              "line": 154,
              "text": "Inspect only the project's seven channels:"
            },
            {
              "line": 158,
              "text": "corp-v1-aerosim-scope"
            },
            {
              "line": 167,
              "text": "Use enough delivery history to avoid duplicate work and only relevant new activity from the other channels. Never inspect another project."
            },
            {
              "line": 169,
              "text": "## Authority and Escalation"
            },
            {
              "line": 173,
              "text": "- run application unit tests, builds, and read-only checks; non-unit suites run through UAT;"
            },
            {
              "line": 174,
              "text": "- repair clear CI/build defects within authorized scope;"
            },
            {
              "line": 175,
              "text": "- add unit tests for an approved behavior and request non-unit coverage from UAT;"
            },
            {
              "line": 178,
              "text": "- reactivate only the disabled `jarvis-at-skic` Gitea account under the standing self-recovery contract in [`references/runtime-and-ci.md`](references/runtime-and-ci.md)."
            },
            {
              "line": 180,
              "text": "Requires approval before:"
            },
            {
              "line": 182,
              "text": "- changing Scope, UI/UX, requirements, or Architecture to make tests pass without the corresponding primary owner or RootAtSkic as backup;"
            },
            {
              "line": 183,
              "text": "- initiating a protected merge that is outside the exact work already authorized by RootAtSkic or the approved Delivery wave; existing exact-scope authorization remains valid through review, merge, and required post-merge verification unless revoked, narrowed, or expanded into a new high-impact scope;"
            },
            {
              "line": 188,
              "text": "## Evidence Requirements"
            },
            {
              "line": 190,
              "text": "Internally retain repository, branch, pull request, exact commit identity, exact PR head and reviewed revision, Hermes main-session review verdict, changed paths, commands, local results, CI run/task IDs and status, and relevant requirement/ADR references. In user-facing updates, lead with the linked branch and PR, exact Task status, active gate code and title, exact failed file/test/job when applicable, and the next action. Omit raw commit identities and digests by default unless RootAtSkic requests audit detail or an identity mismatch is itself the blocker. Report blockers honestly."
            }
          ]
        },
        {
          "path": "references/common-pitfalls.md",
          "installed_sha256": "06d80ca0d25b3d09fbf7ef3debe4f33c559a39445f54b8d7173c4e63614ce0d8",
          "lines": 29,
          "remote_sha256": "06d80ca0d25b3d09fbf7ef3debe4f33c559a39445f54b8d7173c4e63614ce0d8",
          "matches_remote": true,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/corp-v1-channel-delivery-aerosim/src/commit/0c0e68bde3189fd96415af6035200cf7fcb9c877/references/common-pitfalls.md",
          "excerpts": [
            {
              "line": 1,
              "text": "# AeroSim Delivery Pitfalls"
            },
            {
              "line": 3,
              "text": "Load this reference when a gate fails, when work resumes after interruption, or before reporting a blocker, CI result, merge, or completion."
            },
            {
              "line": 5,
              "text": "## Common Pitfalls"
            },
            {
              "line": 7,
              "text": "1. Patching symptoms before understanding root cause."
            },
            {
              "line": 19,
              "text": "13. Treating an ideation-board entry, generated presentation, or broad approval as equivalent to an exact Task `READY_FOR_DELIVERY` transition\u2014or requiring a second admission after that verified transition."
            },
            {
              "line": 21,
              "text": "15. Continuing to report or validate an open PR head after the PR merged during the run; re-resolve the merge/default-branch SHA and require local validation plus CI for that exact merge commit."
            },
            {
              "line": 23,
              "text": "17. Editing in a local-only worktree or on a differently named local branch before the exact remote branch and PR head are visible and verified."
            },
            {
              "line": 24,
              "text": "18. Starting `pnpm`, tests, or a background worker before checking the toolchain and required environment inside that exact execution context."
            },
            {
              "line": 25,
              "text": "19. Reporting only \u201cfailed\u201d or leading with hashes instead of the linked branch/PR and the exact failed command, file, test, or CI job."
            },
            {
              "line": 26,
              "text": "20. Stopping at a repairable gate failure, abandoning the active Task, or starting another Task instead of fixing forward and rerunning the gate until it passes."
            },
            {
              "line": 27,
              "text": "21. Escalating a runner restart for Delivery CI stalled while repeatedly downloading stable build/toolchain prerequisites before checking AeroSim's base-image repository. Browser and Playwright image preparation belongs to UAT; Delivery routes that need rather than adding non-unit infrastructure to application CI."
            },
            {
              "line": 28,
              "text": "22. Treating \u201cthe documentation lock is unavailable\u201d as proof that this Task owns it. After starting a background `flock -n` holder, immediately verify that exact holder process remains running and acquired the lock; an exited holder with status 1 means another Task owns the lock. Wait for that owner to finish, then acquire and verify this Task's own holder before refresh, rebuild, validation, review/CI, merge, and remote readback."
            },
            {
              "line": 29,
              "text": "23. Requiring a second merge authorization after RootAtSkic already authorized completing the exact end-to-end change or approved wave. Existing exact-scope authorization remains valid through review, merge, and required post-merge verification unless it is revoked, narrowed, or a new high-impact scope appears."
            }
          ]
        },
        {
          "path": "references/failure-and-documentation.md",
          "installed_sha256": "b225ee5fd0fb57571cab0e388399a1c1235164c606f56cc01c30982f2d750fd0",
          "lines": 64,
          "remote_sha256": "720d6823006b6e08c80188efa6d8d2d0567c622eea37372ca4996ce1fa7a9749",
          "matches_remote": false,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/corp-v1-channel-delivery-aerosim/src/commit/0c0e68bde3189fd96415af6035200cf7fcb9c877/references/failure-and-documentation.md",
          "excerpts": [
            {
              "line": 1,
              "text": "# AeroSim Failure and Documentation Operations"
            },
            {
              "line": 3,
              "text": "Load this reference when repairing a failure or changing Delivery-owned project documentation."
            },
            {
              "line": 5,
              "text": "## Failure-Repair Workflow"
            },
            {
              "line": 9,
              "text": "1. verify repository, linked branch and PR, exact failed command/file/test/job, failure class, and exact logs;"
            },
            {
              "line": 11,
              "text": "3. determine root cause before editing;"
            },
            {
              "line": 17,
              "text": "9. inspect CI for the exact new head SHA;"
            },
            {
              "line": 21,
              "text": "The skill must continue with a safe forward fix when the failure and expected behavior are clear. A repairable failure does not end the Task or authorize moving to another gate or Task. If evidence is insufficient, requirements/architecture conflict, an external dependency is unavailable, or the required action exceeds authority, keep the Task in the same gate, record the separate blocker with evidence and owner, and raise the issue in the relevant channel:"
            },
            {
              "line": 26,
              "text": "- implementation-only blocker \u2192 `delivery`."
            },
            {
              "line": 30,
              "text": "## Documentation Ownership"
            },
            {
              "line": 34,
              "text": "### Ways of Working \u2014 five sections"
            },
            {
              "line": 38,
              "text": "3. **Planning and ownership** \u2014 backlog, sequencing, owners, dependencies, blockers, and escalation."
            },
            {
              "line": 42,
              "text": "### Engineering \u2014 five sections"
            },
            {
              "line": 45,
              "text": "2. **Development and unit testing** \u2014 commands, unit-test fixtures, debugging, and the handoff to UAT for every non-unit suite."
            },
            {
              "line": 47,
              "text": "4. **Packaging and deployment preparation** \u2014 containers, Helm/GitOps interfaces, configuration, and release inputs."
            },
            {
              "line": 48,
              "text": "5. **Operations and troubleshooting** \u2014 observability, known failure modes, recovery checks, and evidence collection."
            },
            {
              "line": 50,
              "text": "Update these pages from verified project practice. Do not document a planned process as already operational."
            },
            {
              "line": 52,
              "text": "## Synchronization Workflow"
            },
            {
              "line": 54,
              "text": "When planning documentation, Architecture structure, or Kanban state changes while implementation remains gated, follow [`planning-artifact-reconciliation.md`](planning-artifact-reconciliation.md). It defines fresh remote-head discovery, exact-merge-SHA validation, `READY_FOR_DELIVERY` handoff interpretation, evidence boundaries, and duplicate-status suppression."
            },
            {
              "line": 57,
              "text": "2. Before claiming new task work, verify exact-item `Approved for Implementation` evidence, exact Task `READY_FOR_DELIVERY` state, task dependencies, approved UI/UX handoff when applicable, and target version."
            },
            {
              "line": 60,
              "text": "5. Complete at least one useful safe activity when an authorized path exists; otherwise document and route the exact blocker."
            },
            {
              "line": 61,
              "text": "6. Escalate uncertainty or contradiction rather than selecting direction."
            }
          ]
        },
        {
          "path": "references/gitea-actions-policy.md",
          "installed_sha256": "b008022008643333cd69e35f02aa9afd3ed3c5df4aa477996e3d7be053d89a6b",
          "lines": 68,
          "remote_sha256": "b072049884e47214663b8a65f8a7b3b96bbaf21e6b76c885ba335bb45b7b5ac9",
          "matches_remote": false,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/corp-v1-channel-delivery-aerosim/src/commit/0c0e68bde3189fd96415af6035200cf7fcb9c877/references/gitea-actions-policy.md",
          "excerpts": [
            {
              "line": 1,
              "text": "# AeroSim Gitea Actions policy"
            },
            {
              "line": 3,
              "text": "Use this policy for `corp-v1-aerosim/corp-v1-aerosim`. The repository is a pnpm/Turborepo monorepo. Keep the workflow model simple: every validation build covers all applications; do not add affected-workspace selection."
            },
            {
              "line": 5,
              "text": "## 1. Non-default branch without a pull request"
            },
            {
              "line": 9,
              "text": "\u2514\u2500\u2500 branch workflow run for the exact commit"
            },
            {
              "line": 15,
              "text": "        \u251c\u2500\u2500 service-free fast unit tests"
            },
            {
              "line": 21,
              "text": "## 2. Non-default branch with a pull request"
            },
            {
              "line": 26,
              "text": "\u2514\u2500\u2500 pull-request workflow run for the exact PR head"
            },
            {
              "line": 31,
              "text": "    \u2502   \u2514\u2500\u2500 service-free unit tests"
            },
            {
              "line": 34,
              "text": "    \u251c\u2500\u2500 lint and package the Helm chart locally"
            },
            {
              "line": 38,
              "text": "The local images and packaged chart are disposable validation outputs proving buildability and packageability. The PR workflow must use no Harbor or registry credentials and must never authenticate, push, publish, deploy, or mutate GitOps. Do not start PostgreSQL or application services. Do not run journeys or any other non-unit suite. A newer push cancels the superseded PR run."
            },
            {
              "line": 40,
              "text": "## 3. Pull request merged into `test`"
            },
            {
              "line": 43,
              "text": "push to test at the exact integrated commit"
            },
            {
              "line": 47,
              "text": "\u251c\u2500\u2500 package the Helm chart with exact source/image identity"
            },
            {
              "line": 48,
              "text": "\u251c\u2500\u2500 publish the immutable OCI Helm chart to Harbor"
            },
            {
              "line": 52,
              "text": "Do not start PostgreSQL or the deployed applications. Do not run journeys or long-flight UAT. Do not deploy from Gitea Actions; GitOps owns deployment."
            },
            {
              "line": 54,
              "text": "## Common rules"
            },
            {
              "line": 56,
              "text": "- Use the exact workflow display names `CI / Branch Readiness`, `CI / Pull Request Validation`, and `CD / Test Artifact Publication` for the three workflow types."
            },
            {
              "line": 57,
              "text": "- Workflows must check out and validate the exact event commit."
            },
            {
              "line": 59,
              "text": "- Feature/PR validation uses no Harbor credentials; it must build the candidate images and lint/package the candidate Helm chart locally without pushing or publishing anything."
            },
            {
              "line": 60,
              "text": "- Only the exact `test` integration workflow may authenticate to Harbor and publish images and the chart."
            },
            {
              "line": 61,
              "text": "- Full-SHA image tags must never overwrite different content. A retry may resume an existing tag only when its remote OCI config digest exactly matches the validated local image ID."
            },
            {
              "line": 62,
              "text": "- Every `test` commit gets a source-specific immutable chart version containing the full 40-character commit SHA. A retry may resume an existing chart only when its pulled package bytes exactly match the deterministic local package; different content must fail closed."
            },
            {
              "line": 63,
              "text": "- Validate bounded image-transfer chunks, checksums, loaded-image provenance, canonical source URL/timestamp, and exact image/chart evidence shape before credentials or combined evidence are accepted. Preserve the canonical source URL and publication timestamp in the final aggregate evidence."
            },
            {
              "line": 64,
              "text": "- The publication result must run with `always()` and fail unless both image and chart publication jobs succeeded."
            },
            {
              "line": 65,
              "text": "- Build all applications on every validation run; no changed-workspace optimization."
            },
            {
              "line": 66,
              "text": "- All non-unit tests\u2014including every browser journey, integration suite, security-boundary check, container-runtime test, and Helm-rendering test\u2014belong in `corp-v1-aerosim/corp-v1-aerosim-e2e`; application Actions run unit tests only."
            },
            {
              "line": 68,
              "text": "- Preserve one clear required PR result and one clear publication result."
            }
          ]
        },
        {
          "path": "references/planning-artifact-reconciliation.md",
          "installed_sha256": "e33e2395f10b236726bb2c361ad953d8c36892ea62165aaf9db75cb830452103",
          "lines": 38,
          "remote_sha256": "e33e2395f10b236726bb2c361ad953d8c36892ea62165aaf9db75cb830452103",
          "matches_remote": true,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/corp-v1-channel-delivery-aerosim/src/commit/0c0e68bde3189fd96415af6035200cf7fcb9c877/references/planning-artifact-reconciliation.md",
          "excerpts": [
            {
              "line": 1,
              "text": "# Planning-artifact reconciliation"
            },
            {
              "line": 3,
              "text": "Use this procedure when Scope, Architecture, Kanban, Ways of Working, Engineering, or release-readiness documentation changes while Feature implementation remains gated."
            },
            {
              "line": 5,
              "text": "## Procedure"
            },
            {
              "line": 10,
              "text": "4. Resolve the current source ref and full head SHA from the fresh PR API response before fetching. Do not infer the branch name from a PR title, an older checkout, a prior report, or a similarly named local branch. Fetch that exact `head.ref` into an unambiguous temporary remote-tracking ref, create a clean detached worktree from it, and assert `git rev-parse HEAD` equals the API-reported full SHA before validation."
            },
            {
              "line": 11,
              "text": "5. Validate the exact head in that clean or detached worktree using the repository's own validation command. Explicitly execute package-manager commands with the worktree as the process working directory (or begin the shell block with `cd \"$worktree\"`); setting up the worktree in an earlier command does not make it the current directory for a later process. Install dependencies with the repository's declared frozen-lockfile command when the clean worktree has no installed dependencies. If the package manager shim is unavailable, use the repository-declared package-manager launcher rather than changing project files. Before running a composite package script, verify that child scripts can resolve the package-manager executable too: `corepack pnpm validate` can start successfully yet fail when `validate` invokes nested `pnpm ...` commands. In that case, create a writable temporary shim directory with `corepack enable --install-directory <cache>/pnpm-bin`, prepend it to `PATH`, and run the declared command (for example, `PATH=<cache>/pnpm-bin:$PATH pnpm validate`). Keep the shim and dependency cache outside committed content, run `git diff --check`, and confirm the worktree remains clean afterward."
            },
            {
              "line": 12,
              "text": "6. Query Gitea Actions tasks and match the full exact head SHA. Require each matching task to report `success`; do not transfer a PR-head result to the eventual merge head. Keep Gitea identifiers distinct: the Actions task API `id`, its `run_number`, and any UI run URL/ID are separate fields and may not have the same value. Report the exact field actually returned by the verified endpoint (for example, `task id 684, run_number 66`) rather than relabeling a channel-reported UI run ID as a task ID."
            },
            {
              "line": 13,
              "text": "7. Read back the changed planning artifacts through the authenticated raw-file API at the exact PR source ref and verify their **semantics**, not merely keyword presence. Headings and lifecycle guidance naturally contain terms such as `READY_FOR_DELIVERY`, so keyword presence alone does not prove a Task is pickup-ready. For Kanban, assert the exact Task row and status, the evidence timestamp, and any newly cited head/task pairs; also assert superseded pairs are absent when the change claims reconciliation. Parse each lifecycle section independently and verify the Task's actual row rather than using loose whole-file heuristics such as a state-name count. If an automated semantic probe disagrees with the rendered Markdown, read the relevant numbered lines and correct the probe before reporting."
            },
            {
              "line": 14,
              "text": "8. Immediately before reporting, re-fetch the PR, its remote source ref, the default branch, and exact-SHA tasks. Scheduled writers or human reviewers can change repository state during local validation."
            },
            {
              "line": 16,
              "text": "   - If the PR merged or closed during validation, stop describing it as open. Fetch the freshly resolved merge/default-branch SHA, create or switch to a clean detached worktree at that exact commit, rerun the repository's declared validation and semantic probes there, and require a successful Actions task whose full `head_sha` equals the merge commit. PR-head CI must not be transferred to the merge head."
            },
            {
              "line": 17,
              "text": "   - Re-fetch once more after merge-head validation and assert the PR state, default-branch SHA, local validated SHA, and latest exact-SHA terminal task still agree. If another Delivery report already covers that identical evidence, remain silent."
            },
            {
              "line": 20,
              "text": "   - An exact Task is Focus-admitted and pickup-ready when its status is `READY_FOR_DELIVERY`; there is no second Focus-admission gate."
            },
            {
              "line": 21,
              "text": "   - A newly added Kanban page or merged board does not itself move an exact Task to `READY_FOR_DELIVERY`."
            },
            {
              "line": 22,
              "text": "10. Re-evaluate every implementation gate from the source files and approvals: Approved for Implementation, FR/NFR and ADR context, task/dependency plan, target version, acceptance evidence, and exact Task `READY_FOR_DELIVERY` state."
            },
            {
              "line": 23,
              "text": "11. If the gate remains closed, do not create application changes. A useful delivery activity can be exact-head build/CI verification, durable documentation reconciliation, or a newly evidenced blocker report."
            },
            {
              "line": 24,
              "text": "12. Report only deltas not already covered by the latest delivery message. Include repository, branch, full SHA, local commands/results, exact-SHA task IDs, and the specific missing gate."
            },
            {
              "line": 26,
              "text": "## Evidence interpretation"
            },
            {
              "line": 32,
              "text": "| Kanban Board lists an item in ideation | The unfinished scope record is visible | Exact Task `READY_FOR_DELIVERY` handoff |"
            },
            {
              "line": 36,
              "text": "## Avoid duplicate status"
            },
            {
              "line": 38,
              "text": "Before posting, compare the proposed report with enough Delivery history to find the newest marker, head SHA, CI task, and blocker set. If there is no new human decision, repository head, CI result, actionable failure, or durable documentation change, return the configured silent response instead of restating the same gate."
            }
          ]
        },
        {
          "path": "references/repository-entry-and-visibility.md",
          "installed_sha256": "31f2ccf03cc9bf7270a25aa930f0099e6b7d9272a51860d9b7b493aad0eed98f",
          "lines": 60,
          "remote_sha256": "5d37a6c3dda1eb0057787c645d949c74a60cec7e9aea83380c31927ad7842535",
          "matches_remote": false,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/corp-v1-channel-delivery-aerosim/src/commit/0c0e68bde3189fd96415af6035200cf7fcb9c877/references/repository-entry-and-visibility.md",
          "excerpts": [
            {
              "line": 1,
              "text": "# AeroSim Repository Entry and Visibility"
            },
            {
              "line": 3,
              "text": "Load this reference before claiming a Task, creating its application branch, opening its PR, or resuming its worktree."
            },
            {
              "line": 5,
              "text": "All local repository entry happens inside `corp-v1-aerosim-delivery/workspace/<TASK_ID>/`. During preparation, create only the required repository clones: `application/` and `documentation/`. Optional Task-local `review/`, `cache/`, `tmp/`, `logs/`, or `artifacts/` folders may be added later only when a concrete Task activity needs them. Reusable installed toolchains and technical dependencies are the sole exception: place them in `corp-v1-aerosim-delivery/cache/`, never in a Task workspace. A repository branch, clone, Task evidence file, or other Task-local support folder outside the selected Task directory does not satisfy this gate."
            },
            {
              "line": 7,
              "text": "## Implementation Entry Gate"
            },
            {
              "line": 9,
              "text": "Before claiming a Task for Feature implementation, verify in project documentation:"
            },
            {
              "line": 12,
              "text": "- exact `Approved for Implementation` evidence from the primary Architecture owner `<@1543940246071410706>` or RootAtSkic as backup;"
            },
            {
              "line": 14,
              "text": "- task breakdown and dependencies;"
            },
            {
              "line": 17,
              "text": "- exact Task `READY_FOR_DELIVERY` evidence, which itself records Focus admission by RootAtSkic; and"
            },
            {
              "line": 18,
              "text": "- an approved UI/UX package from Dovydas or RootAtSkic as backup when the task changes user-facing design."
            },
            {
              "line": 20,
              "text": "Delivery must verify the Feature decision and exact Task `READY_FOR_DELIVERY` state before initial coding, then claim the Task by moving it directly to `IN_PROGRESS`. For release correction work, an exact human-authored `READY_FOR_DELIVERY` event that explicitly approves the current release wave and says no second admission gate is required is the governing implementation approval; do not block it merely because the parent Feature retains its historical portfolio status. Continue claimed work in `IN_PROGRESS` while retaining the original handoff evidence. Do not require a second Focus admission after `READY_FOR_DELIVERY` or infer missing evidence from broad delegation, silence, CI, a Board wave, or a planning status. When neither Feature approval nor an explicit human-authored release-wave approval exists, route the exact owning action to Scope, Architecture, or Kanban rather than creating an out-of-scope lifecycle or flow mutation inside an implementation change."
            },
            {
              "line": 22,
              "text": "## Immediate Wave-Start Visibility Gate"
            },
            {
              "line": 24,
              "text": "Before any implementation or repository content change begins on a newly claimed AeroSim Task, establish and verify all of these Gitea checkpoints:"
            },
            {
              "line": 26,
              "text": "1. Create the dedicated local branch with the required Task branch name, create an empty kickoff commit if needed, and push that exact branch immediately."
            },
            {
              "line": 27,
              "text": "2. Open a visible pull request from that exact branch to `test` before editing a repository file. Do not wait for implementation, tests, review, or a finished diff."
            },
            {
              "line": 28,
              "text": "3. Put the Task ID, wave, planned scope, and current Task state in the PR so RootAtSkic and the team can monitor progress from the beginning."
            },
            {
              "line": 29,
              "text": "4. Immediately write the exact branch name and clickable PR URL back to the matching [AeroSim Delivery](https://jarvis-jr-v1.lego-cloud.eu/kanban?board=aerosim-uat-gitops) card before any test, source, configuration, documentation, diagnosis-by-edit, or other repository content change. Populate the card's native branch field when supported; otherwise add a card comment. A PR that exists in Gitea but is not linked from its active card does not satisfy visibility."
            },
            {
              "line": 30,
              "text": "5. Read the PR and remote ref back, then prove that the local branch name exactly equals the remote branch and PR head ref, the local kickoff commit equals the remote branch head, and the PR base is `test`."
            },
            {
              "line": 31,
              "text": "6. Only after the remote branch, visible pull request, and card link already exist and all identities match may Delivery edit tests, source, configuration, documentation, or other repository content."
            },
            {
              "line": 33,
              "text": "No repository file may be edited before this gate passes. A differently named local convenience branch that merely tracks the PR branch does not pass, because it hides the branch identity during status reporting and makes accidental publication to the wrong ref easier. When resuming an existing worktree, fetch the PR head explicitly, require the current local branch name to equal the PR head ref, and fast-forward before editing; if the names differ, repair the local branch identity without force-pushing or creating a competing PR."
            },
            {
              "line": 35,
              "text": "One Task requires one dedicated branch and one visible PR unless AeroSim Architecture or RootAtSkic explicitly approves a different grouping. Do not hide several independently deliverable Tasks behind a shared wave branch. If Gitea refuses a no-diff PR, create and push an empty kickoff commit\u2014never a filler file change\u2014then open the PR. A locally created worktree, unpushed branch, compare URL, or push without PR readback does not satisfy this gate."
            },
            {
              "line": 37,
              "text": "### Branch and pull-request naming contract"
            },
            {
              "line": 39,
              "text": "For every AeroSim delivery Task, use this exact source-branch shape:"
            },
            {
              "line": 51,
              "text": "- `<release>` is the Task's exact target release, without a leading `v`. Do not derive it from a package manifest, invent it, or replace it with `unreleased`, `next`, or a wave label."
            },
            {
              "line": 52,
              "text": "- `<TASK_ID>` is the exact Task ID, preserving its uppercase spelling."
            },
            {
              "line": 53,
              "text": "- `<short-explanation>` is a concise lowercase ASCII kebab-case description (`a-z`, `0-9`, and single hyphens only)."
            },
            {
              "line": 54,
              "text": "- `<attempt-digit>` is an unpadded positive decimal attempt number. Start at `1`; increment it only when a new branch/PR attempt is required. Never reuse an attempt number for the same release, Task, and explanation."
            },
            {
              "line": 55,
              "text": "- The complete branch must pass `git check-ref-format --branch`, contain exactly one `/`, and remain a valid Gitea branch ref. Spaces, repeated separators, `..`, `@{`, backslash, control characters, and the Git-forbidden characters `~ ^ : ? * [` are not allowed."
            },
            {
              "line": 56,
              "text": "- Resolve collisions against freshly fetched local and remote refs before creating the branch. Never force-push or repurpose an earlier attempt."
            },
            {
              "line": 58,
              "text": "The Gitea pull-request title must exactly equal the source branch name. Gitea's PR API treats the title as a string rather than a Git ref, but using the exact branch value provides one unambiguous release/Task/attempt identity across the branch list, PR list, API, CI, and evidence records. Do not prepend `feat:`, `fix:`, an emoji, a wave label, or other prose to the title."
            },
            {
              "line": 60,
              "text": "Before creating the PR, verify the Gitea technical prerequisites: the source branch is pushed, the base branch is exactly `test`, head and base differ, the branch contains a commit not already in `test`, and no open PR already exists for the same head/base pair. Create the PR with explicit `title`, `head`, `base`, and `body` fields. The body must record the Task ID, release, attempt, wave, planned scope, Task state, validation plan, and requirement/ADR links. Read the PR back and require exact title, head ref/SHA, and `test` base; a compare URL or successful API response alone is not evidence that the PR contract is satisfied."
            }
          ]
        },
        {
          "path": "references/runtime-and-ci.md",
          "installed_sha256": "c681d6e6ef394279bdaf7f101d58f7ff328e6423d09b743b6bd1255de52d55a2",
          "lines": 155,
          "remote_sha256": "c681d6e6ef394279bdaf7f101d58f7ff328e6423d09b743b6bd1255de52d55a2",
          "matches_remote": true,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/corp-v1-channel-delivery-aerosim/src/commit/0c0e68bde3189fd96415af6035200cf7fcb9c877/references/runtime-and-ci.md",
          "excerpts": [
            {
              "line": 1,
              "text": "# AeroSim Runtime and CI"
            },
            {
              "line": 3,
              "text": "Load this reference before login, toolchain execution, CI work, image publication, or Gondor resource work."
            },
            {
              "line": 5,
              "text": "## Channel-owned workspace"
            },
            {
              "line": 17,
              "text": "        \u251c\u2500\u2500 review/            # optional: only when separate review files are needed"
            },
            {
              "line": 18,
              "text": "        \u251c\u2500\u2500 cache/             # optional: only when a Task-specific cache is needed"
            },
            {
              "line": 19,
              "text": "        \u251c\u2500\u2500 tmp/               # optional: only when temporary Task files are needed"
            },
            {
              "line": 20,
              "text": "        \u251c\u2500\u2500 logs/              # optional: only when logs must be retained"
            },
            {
              "line": 21,
              "text": "        \u2514\u2500\u2500 artifacts/         # optional: only when outputs must be retained"
            },
            {
              "line": 24,
              "text": "- `cache/` contains reusable technical dependencies installed by Delivery, including pinned Node/pnpm toolchains, Nginx binaries, and package-manager download caches. Browser/Playwright dependencies for non-unit suites belong to the UAT workspace and E2E repository. Reuse verified entries across Tasks instead of reinstalling them. Version or digest cache paths when compatibility matters, verify the resolved executable and version in every execution context, and never store credentials, Task evidence, repository content, or mutable application state there."
            },
            {
              "line": 27,
              "text": "- During `GATE_001_PREP`, create only the required repository clones: `application/` and `documentation/`. If a Task genuinely needs another repository, clone it as another clearly named child."
            },
            {
              "line": 28,
              "text": "- Do not create Task-local `review/`, `cache/`, `tmp/`, `logs/`, or `artifacts/` speculatively. Create each folder only when a concrete activity for that Task needs it."
            },
            {
              "line": 29,
              "text": "- Use Task-local `review/` only for a separate review checkout or review output. Use Task-local `cache/` only for non-reusable data bound to that Task; reusable installed technical dependencies belong in the channel-level `cache/`. Use `tmp/`, `logs/`, and `artifacts/` only for their named Task-local purpose."
            },
            {
              "line": 30,
              "text": "- Run dependency installation, generation, tests, and builds inside the applicable repository clone using that repository's normal paths. Copy or export outputs to `artifacts/` only when they must be retained separately."
            },
            {
              "line": 32,
              "text": "- Before every local repository command, resolve the real working directory and require it to be inside the selected Task directory. A command running elsewhere is invalid evidence and must stop before mutation."
            },
            {
              "line": 33,
              "text": "- Do not reuse another Task's clone or support folder as a shortcut. Resume only from the selected Task's own workspace after fetching and matching its visible PR head."
            },
            {
              "line": 35,
              "text": "## Shared Corp v1 System Login"
            },
            {
              "line": 37,
              "text": "When an authorized task requires login to a Corp v1 system being built or operated, follow the shared policy in `corp-v1-main` and use only the Bitwarden-injected runtime secrets named:"
            },
            {
              "line": 44,
              "text": "Secret availability is capability, not authorization. Verify the destination origin and task purpose before login. Never print, inspect, log, hash, serialize, paste, screenshot, or persist either value; never place a value in a command line, URL, file, repository, prompt, Discord message, browser console, test fixture, CI output, or generated artifact. Never ask a human to paste a value into chat. If a variable is unavailable, report only its missing name and request Bitwarden/gateway injection. Login does not authorize account recovery, MFA or credential changes, permission changes, billing, spending, destructive operations, or access outside the approved project task."
            },
            {
              "line": 46,
              "text": "## Standing Gitea Self-Reactivation"
            },
            {
              "line": 48,
              "text": "RootAtSkic grants Hermes main session in AeroSim Delivery standing authority to reactivate only its own Gitea account, `jarvis-at-skic`, when disablement blocks otherwise authorized AeroSim Delivery work. This is separate project authority; it does not derive from possession of login secrets. When every precondition below passes, Hermes must recover the account and resume the blocked gate without asking for additional per-incident approval."
            },
            {
              "line": 53,
              "text": "2. The original Gitea token identifies, or was previously verified to identify, exactly `jarvis-at-skic`."
            },
            {
              "line": 54,
              "text": "3. Independent preflight proves account disablement: the API rejects that original identity because it is inactive or prohibited, and SSH reports that the account is disabled. A generic `401`, `403`, missing scope, repository denial, expired credential, network failure, or key failure is not enough by itself."
            },
            {
              "line": 55,
              "text": "4. A control-plane read identifies exactly one `jarvis-at-skic` record and confirms `is_active=false` with `prohibit_login=false`."
            },
            {
              "line": 59,
              "text": "1. Load `self-hosted-service-account-recovery` and its `references/gitea-on-truenas-reactivation.md` procedure, plus `home-v1--truenas`, before mutation."
            },
            {
              "line": 61,
              "text": "3. Use an exact, conditional, idempotent mutation that sets only `is_active=true` for `name='jarvis-at-skic'`; require exactly one affected row. Do not change `prohibit_login`, password, MFA, tokens, SSH keys, administrator status, organization/team/repository permissions, or any other account."
            },
            {
              "line": 62,
              "text": "4. Re-read the exact control-plane record in a fresh session and require `is_active=true` and `prohibit_login=false`."
            },
            {
              "line": 63,
              "text": "5. Verify the original token with `GET /api/v1/user` and require `login=jarvis-at-skic`, `active=true`, and `prohibit_login=false`; then verify the original SSH key independently with read-only `git ls-remote` against an authorized AeroSim repository."
            },
            {
              "line": 64,
              "text": "6. Resume the same Delivery gate only after all three verification layers pass. Report the recovery and access paths without exposing credentials."
            },
            {
              "line": 73,
              "text": "The helper is hard-locked to `jarvis-at-skic`, requires separate write and exact-account interlocks, resolves the current Gitea PostgreSQL container, uses a short-lived TrueNAS shell token, updates only an exact inactive non-prohibited row, and requires `UPDATE 1`. The API-key connection must always use `wss://`; TrueNAS automatically revokes an API key presented over plaintext `ws://`. `--allow-plaintext-shell` applies only to the generated 60-second single-use shell token and never permits changing `HL_V1_TRUENAS_URL` to `http://`. If `https://nas.lego-cloud.eu/websocket/shell` times out and its plaintext form redirects with HTTP 301, use the documented direct appliance endpoint `wss://192.168.1.218/api/current` with both `--insecure-truenas` and `--allow-plaintext-shell`: the API key remains encrypted under the bounded self-signed-certificate exception, while only the short-lived token uses the internal plaintext shell route. Run `terminal(command=\"uv run scripts/test_gitea_account_recovery.py\", workdir=\"<skill-repository>\", timeout=90)` after any helper or policy change."
            },
            {
              "line": 75,
              "text": "Stop without mutation and report the exact blocker if the account is not exactly `jarvis-at-skic`, the pre-state is ambiguous, `prohibit_login=true`, the supported contract cannot be verified, the mutation would affect zero or multiple records, or recovery would require changing any excluded security or permission field. This standing authority never permits disabling an account, modifying another account, borrowing another identity, rotating credentials, or broadening access."
            },
            {
              "line": 77,
              "text": "## Delivery Responsibilities"
            },
            {
              "line": 79,
              "text": "### Coding"
            },
            {
              "line": 81,
              "text": "- implement only human-approved Features and their authorized tasks;"
            },
            {
              "line": 82,
              "text": "- load applicable project engineering skills before changing code;"
            },
            {
              "line": 88,
              "text": "### Unit testing"
            },
            {
              "line": 91,
              "text": "- reproduce defects before fixing when practical;"
            },
            {
              "line": 94,
              "text": "- record exact commands and real results."
            },
            {
              "line": 96,
              "text": "### Build and continuous integration"
            },
            {
              "line": 101,
              "text": "- unit-test failures and UAT-reported non-unit failures;"
            },
            {
              "line": 110,
              "text": "### Execution-context toolchain preflight"
            },
            {
              "line": 112,
              "text": "Treat every foreground shell, detached worktree, background process, Hermes main-session review command, and CI job as a separate execution context. Before any repository command or operation runs in an execution context, complete the applicable preflight for that exact context. For a Node/pnpm repository\u2014or any command that may invoke Node or pnpm\u2014the following checks are mandatory:"
            },
            {
              "line": 115,
              "text": "2. run `command -v node`, `node --version`, `command -v pnpm`, and `pnpm --version` inside the exact context, then compare the observed versions with the committed required versions;"
            },
            {
              "line": 118,
              "text": "5. start the repository command only after every applicable preflight check passes."
            },
            {
              "line": 120,
              "text": "Never assume a background process inherits `PATH`, shell activation, Devbox state, Corepack activation, database variables, or other exports from a foreground terminal. A missing `pnpm` executable is always an execution-context/toolchain failure, never an application-code or test failure. Repair the exact context by providing the required toolchain path or activation, repeat the complete preflight there, and rerun the original command before making any claim about repository code."
            },
            {
              "line": 122,
              "text": "### Immutable application images and Helm charts"
            },
            {
              "line": 124,
              "text": "Publish both runtime images and the deployable Helm chart to the existing Harbor registry after exact-head validation on the trusted `test` branch."
            },
            {
              "line": 126,
              "text": "- Package the chart from the exact validated application revision and require `helm lint`, deterministic chart metadata readback, and a SemVer-compatible chart version. Keep the four-part product version in `appVersion` and release metadata; do not force it into Helm's package-version field."
            },
            {
              "line": 127,
              "text": "- Push the package as an OCI artifact to the project-scoped Harbor path, using the same narrowly scoped robot credential already authorized for AeroSim artifact publication. Pass the password through standard input and never print it."
            },
            {
              "line": 128,
              "text": "- Use immutable chart versions: serialize publication for the project and require a Harbor immutable-tag rule covering the chart repository. If the source-specific version already exists, pull it and resume only when its archive bytes hash exactly like the deterministic local package; fail closed on any mismatch. Read the artifact back from Harbor, verify chart name/version/application version, and retain the remote OCI manifest digest and archive checksum as separate release evidence."
            },
            {
              "line": 129,
              "text": "- Configure Argo CD to consume the published OCI chart by exact chart version from Harbor. Do not make the deployment controller clone the private application source repository merely to obtain the chart."
            },
            {
              "line": 130,
              "text": "- Keep GitOps values and immutable image digests in the rendered GitOps repository. The GitOps repository selects the exact chart package and images; Harbor stores the deployable chart and image artifacts."
            },
            {
              "line": 131,
              "text": "- Treat chart build, publication, remote readback, GitOps selection, and live Argo reconciliation as separate gates. A successful chart push alone is not deployment success."
            },
            {
              "line": 134,
              "text": "### Per-project reusable CI base images"
            },
            {
              "line": 136,
              "text": "AeroSim owns `corp-v1-aerosim/corp-v1-aerosim-base-images` as the reusable CI-image repository. Use it when stable runtimes or operating-system tools would otherwise be downloaded repeatedly in application jobs."
            },
            {
              "line": 138,
              "text": "- Keep one folder per image, with its Dockerfile, machine-readable version/platform metadata, usage documentation, build helper, and static publication contract. Runtime smoke belongs to UAT."
            },
            {
              "line": 139,
              "text": "- Pin the upstream base by digest and assert exact runtime/tool versions, architecture, numeric non-root identity, writable workspace, CA trust, Git, shell, and other required tools."
            },
            {
              "line": 140,
              "text": "- Build and perform static/unit validation without registry credentials on pull requests. Publish only from the trusted `test` branch using a narrowly scoped Harbor robot secret; UAT validates the published digest at runtime."
            },
            {
              "line": 142,
              "text": "- Extend fail-closed tests to reject broad publication triggers, mutable tags, secret access from PR paths, shell interpolation, wrong digests, and reintroduced runtime setup actions."
            },
            {
              "line": 144,
              "text": "- Treat the first real Gitea build, Harbor publication, digest readback, and consumer workflow at the exact PR-head SHA as separate Delivery layers. Runtime smoke is a separate UAT layer."
            },
            {
              "line": 145,
              "text": "- Repair image or publication failures forward only. Preserve the last known-good digest for consumers until the replacement image and consuming workflow pass exact-head CI."
            },
            {
              "line": 147,
              "text": "### Gondor runtime-resource placement"
            },
            {
              "line": 149,
              "text": "AeroSim CI jobs and developer workstations must not provision PostgreSQL, queues, object stores, caches, brokers, or other application runtime services. Provision Delivery-owned non-production resources on Gondor through the committed `corp-v1-aerosim/gondor-v1-tmpl-aerosim` template rendered into `corp-v1-aerosim/gondor-v1-aerosim-dev`. Production resources and `corp-v1-aerosim/gondor-v1-aerosim` remain outside Delivery authority and are owned by Releases."
            },
            {
              "line": 152,
              "text": "- Delivery CI may orchestrate exact-head unit/build validation, but it must not start application runtime-resource service containers or receive application-resource credentials. Harmless process-local unit-test doubles and build tools are not runtime resources. UAT owns resource-dependent tests and their restricted execution; Delivery only provisions an approved development dependency and hands its exact identity to UAT. Do not expose PostgreSQL or another internal resource through Ingress, NodePort, LoadBalancer, or a runner-accessible public endpoint."
            },
            {
              "line": 153,
              "text": "- Persistent Gondor storage must follow `home-v1--truenas`. For durable AeroSim data, create a purpose-specific TrueNAS dataset and NFS share restricted to Gondor node networks, then bind a static PV/PVC with `persistentVolumeReclaimPolicy: Retain` and `storageClassName: \"\"`. Do not use the default dynamic `truenas-csi` StorageClass for durable data because its `Delete` reclaim policy can remove the backing dataset with the PVC."
            },
            {
              "line": 154,
              "text": "- Use only External Secrets backed by the existing Bitwarden `ClusterSecretStore`; never commit connection strings or credentials."
            },
            {
              "line": 155,
              "text": "- Provision and verify required Gondor resources before deploying dependent AeroSim code. Resource-independent implementation may proceed, but deployment remains blocked until the dependency is healthy and consumed by the application."
            }
          ]
        },
        {
          "path": "references/single-subagent-task-delivery.md",
          "installed_sha256": "78bed415a694049b2a400785cfbf846e569023f43b9afd2328bcc19997a379ab",
          "lines": 121,
          "remote_sha256": "d0c431689d0264a2a267e5bfe7da08e5c4b5513ee23fc472863c7f3649e5c57b",
          "matches_remote": false,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/corp-v1-channel-delivery-aerosim/src/commit/0c0e68bde3189fd96415af6035200cf7fcb9c877/references/single-subagent-task-delivery.md",
          "excerpts": [
            {
              "line": 1,
              "text": "# Single-Subagent Task Delivery"
            },
            {
              "line": 3,
              "text": "Use one isolated `delegate_task` worker per AeroSim Task while the Hermes main session remains the governance and verification boundary. RootAtSkic grants standing authorization to drain independently eligible current-release Focus work concurrently, each Task with its own worker and workspace. Do not wait for a repeated pickup instruction when a Task is exactly `READY_FOR_DELIVERY` with satisfied dependencies, or when an `IN_PROGRESS` Task has lost its executor and has a recoverable durable checkpoint. When several Tasks require human input, discuss their blockers with RootAtSkic one Task at a time; this does not pause independent workers. This mode preserves the six delivery gates, exact Task status flow, repository controls, and dependency order; it changes who performs the mechanical delivery work, not the evidence required."
            },
            {
              "line": 5,
              "text": "## Preconditions"
            },
            {
              "line": 7,
              "text": "Before dispatch, the main session must:"
            },
            {
              "line": 10,
              "text": "2. Read the current documentation `test` branch and verify each selected Task is exactly `READY_FOR_DELIVERY` or is a recoverable `IN_PROGRESS` Task, its applicable implementation approval is present, dependencies are satisfied, and no selected Task depends on another selected or active Task."
            },
            {
              "line": 14,
              "text": "Do not dispatch if scope requires a human decision, missing credentials cannot be recovered under standing authority, or the work cannot be completed without crossing project or channel boundaries."
            },
            {
              "line": 16,
              "text": "## Workspace and Cache Contract"
            },
            {
              "line": 39,
              "text": "Create only directories the Task needs. Shared `cache/` holds reusable Delivery build toolchains, package stores, Nginx binaries, and other pinned build dependencies. Browser, Playwright, PostgreSQL, and other non-unit test dependencies belong to UAT. It must not contain repository worktrees, Task evidence, database data directories, credentials, tokens, or mutable project records. Task-local `tmp/`, `logs/`, and `artifacts/` contain disposable runtime state and exact evidence. A Task-local `cache/` is permitted only for inputs that are unsafe to share across Tasks."
            },
            {
              "line": 41,
              "text": "The worker may consume an existing shared cache. It may add a missing reusable dependency only when the version/source is pinned and the addition is verified; it must report what changed. Never delete or silently replace shared cache entries during Task delivery."
            },
            {
              "line": 43,
              "text": "## Worker Contract"
            },
            {
              "line": 45,
              "text": "Spawn exactly one leaf worker with `delegate_task`. The packet must be self-contained because the worker has no parent-session context. Require the worker to:"
            },
            {
              "line": 48,
              "text": "2. Execute `GATE_001_PREP` through `GATE_005_TASK_POST_MERGE` in order, stopping fail-closed on a blocker."
            },
            {
              "line": 49,
              "text": "3. Claim `READY_FOR_DELIVERY \u2192 IN_PROGRESS` visibly before implementation."
            },
            {
              "line": 50,
              "text": "4. Work only inside `workspace/<TASK_ID>/` and the channel shared cache."
            },
            {
              "line": 51,
              "text": "5. Use TDD for behavior changes, run repository-native quality gates, and capture exact command results."
            },
            {
              "line": 52,
              "text": "6. Open/update PRs, review the complete frozen exact-head diff, merge only where standing project authority permits, verify post-merge CI and immutable publication, and for every runtime-affecting Task complete the mandatory reviewed development GitOps deployment, Argo CD/pod/image/runtime readback, and UAT tuple handoff before recording `TO_BE_RELEASED` evidence. Only a verified non-runtime `not_applicable` reason may skip development deployment."
            },
            {
              "line": 53,
              "text": "7. Never change Scope, requirements, UI/UX, Architecture, release state, secrets, permissions, costs, or another Task to make delivery pass."
            },
            {
              "line": 54,
              "text": "8. Return a structured result with Task/status/gate, workspace, branches, commits, PR URLs/numbers, reviewed exact heads, changed paths, local test counts, CI run IDs/statuses, merge commits, documentation evidence, cache additions, blocker details, and the next eligible Task set."
            },
            {
              "line": 58,
              "text": "## Same-Wave Parallel Dispatch"
            },
            {
              "line": 62,
              "text": "1. Select every independently eligible `READY_FOR_DELIVERY` Task and recover orphaned `IN_PROGRESS` Tasks whose dependencies are satisfied."
            },
            {
              "line": 65,
              "text": "4. If application boundaries overlap semantically, pause the later Task until the earlier merge lands, then refresh, retest, and review the later exact head."
            },
            {
              "line": 66,
              "text": "5. A genuine documented blocker stops only that Task; continue independent workers."
            },
            {
              "line": 67,
              "text": "6. Discuss blockers requiring RootAtSkic input one Task at a time, completing one decision or input request before presenting the next."
            },
            {
              "line": 70,
              "text": "## Main-Session Verification"
            },
            {
              "line": 72,
              "text": "A worker summary is a claim, not proof. Before reporting completion, the main session must independently:"
            },
            {
              "line": 76,
              "text": "3. Query exact CI runs and verify required jobs belong to the reported immutable head or merge revision."
            },
            {
              "line": 77,
              "text": "4. Verify image publication or deployment evidence where the gate requires it."
            },
            {
              "line": 78,
              "text": "5. Inspect the recorded review packet and confirm its PASS applies to the exact merged candidate tree with no later source edit."
            },
            {
              "line": 82,
              "text": "If verification fails, do not repeat the worker's completion claim. Resume the same Task with the same worker only while it is live; otherwise dispatch one replacement worker with the full checkpoint and exact blocker. A replacement is a recovery attempt for the same Task, not a second concurrent worker."
            },
            {
              "line": 84,
              "text": "## Dispatch Template"
            },
            {
              "line": 102,
              "text": "## Failure and Recovery"
            },
            {
              "line": 104,
              "text": "- Worker reports a requirement contradiction or owner decision: mark the Task `BLOCKED` with evidence and stop."
            },
            {
              "line": 105,
              "text": "- Worker exits before durable publication: inspect the workspace and remotes, freeze a checkpoint, and dispatch one replacement for the same Task."
            },
            {
              "line": 106,
              "text": "- Worker changes files after exact-head PASS: invalidate the verdict and rerun review."
            },
            {
              "line": 108,
              "text": "- Remote state differs from the summary: remote and canonical documentation state win; reconcile before continuing."
            },
            {
              "line": 110,
              "text": "## Verification Checklist"
            },
            {
              "line": 112,
              "text": "- [ ] Every active Task is independently eligible in the authorized current wave and has exactly one worker."
            },
            {
              "line": 117,
              "text": "- [ ] Claim is visible before implementation."
            },
            {
              "line": 118,
              "text": "- [ ] Exact-head review and CI identities are recorded."
            },
            {
              "line": 119,
              "text": "- [ ] Main session independently read back every external side effect."
            }
          ]
        },
        {
          "path": "references/status-and-reporting.md",
          "installed_sha256": "1d73ae35a8ac3620bd6b436e699f0e72ea72a5c5dd283dd3386e74609d2f8a55",
          "lines": 81,
          "remote_sha256": "1d73ae35a8ac3620bd6b436e699f0e72ea72a5c5dd283dd3386e74609d2f8a55",
          "matches_remote": true,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/corp-v1-channel-delivery-aerosim/src/commit/0c0e68bde3189fd96415af6035200cf7fcb9c877/references/status-and-reporting.md",
          "excerpts": [
            {
              "line": 1,
              "text": "# AeroSim Status and Reporting"
            },
            {
              "line": 5,
              "text": "## Scope Status Vocabulary"
            },
            {
              "line": 7,
              "text": "Before reporting or changing any lifecycle status, read the project's current `ways-of-working/scope.md`. Use only the values defined there."
            },
            {
              "line": 21,
              "text": "`READY_FOR_DELIVERY` is the exact Task's Focus-admitted, pickup-ready state. It means Architecture, dependencies, implementation boundaries, verification steps, and Kanban selection are complete; implementation has not started. Delivery claims that Task by moving it directly to `IN_PROGRESS`. Never require or invent a second Focus-admission decision after `READY_FOR_DELIVERY`."
            },
            {
              "line": 25,
              "text": "Never invent, abbreviate, or substitute lifecycle values such as `Ready`, `Active`, `Delivered`, `Complete`, `Awaiting CI`, or `Overrun`. Those phrases may describe activity only when clearly separated from the status field. If the status cannot be verified, report `status unverified` and inspect the source record rather than guessing."
            },
            {
              "line": 27,
              "text": "A merge or successful CI does not automatically mean `DONE`. Releasable application Tasks normally move from `IN_PROGRESS` to `TO_BE_RELEASED`; `DONE` requires the applicable release or completion evidence. Status reports must show the exact value even when a separate plain-language action is also included."
            },
            {
              "line": 29,
              "text": "## Wave Status Report"
            },
            {
              "line": 31,
              "text": "When RootAtSkic explicitly asks for project, delivery, blocker, progress, or wave status, begin the status report with this exact first-line structure, with no text before it:"
            },
            {
              "line": 37,
              "text": "Each value must be an actual dependency-wave label calculated by the AeroSim Release Task Board, such as `Wave 11`, `Wave 12`, or `Wave 13`. Never substitute activities, work descriptions, statuses, invented phases, or ad hoc labels. Resolve the current wave and its adjacent existing waves from the Board selector and task assignments; do not infer them from conversation wording. Use `none` only when the Board has no previous or next wave."
            },
            {
              "line": 39,
              "text": "Do not repeat the wave header or task chapter in ordinary acknowledgements, implementation narration, questions, or non-status replies. Use them only when RootAtSkic explicitly asks for status."
            },
            {
              "line": 41,
              "text": "### Current-wave task chapter"
            },
            {
              "line": 46,
              "text": "## Wave XX (Current)"
            },
            {
              "line": 48,
              "text": "- TASK-XXXX \u2014 <status> \u2014 Responsible: <Hermes main session or concrete owning process>"
            },
            {
              "line": 51,
              "text": "Replace `XX` with the current Board wave number. List every Task assigned to that wave\u2014never a sample or only changed Tasks\u2014and use each Task's verified flow status. For responsibility, name Hermes main session when it actively owns the Task; otherwise name the concrete owning process, such as Delivery pickup, implementation/review, integration, or Release handoff. When no active executor or owning process exists, say `Unassigned`; never invent an agent, process, or ownership. Refresh the Board/task records and live process state before reporting when they may have changed."
            },
            {
              "line": 53,
              "text": "### Human-facing delivery updates"
            },
            {
              "line": 55,
              "text": "In every implementation, review, CI, blocker, merge, handoff, or completion update, lead with the linked branch and pull request where the work is visible. The next line must identify the Task, exact Task status, active gate code, and gate title:"
            },
            {
              "line": 79,
              "text": "Do not invent a separate phase or shorthand. Do not lead with or routinely include commit SHAs, tree hashes, digests, internal worktree paths, or temporary review-commit identities; retain those for automated verification and provide them only when RootAtSkic asks for audit detail or when an identity mismatch itself is the blocker."
            },
            {
              "line": 81,
              "text": "When anything fails, identify the exact failed command, file, test, or CI job. State its failure classification\u2014application code, test assertion, workflow, infrastructure, credentials, or local execution environment\u2014then provide the expected behavior and observed behavior, whether repository or remote state changed, and the concrete repair or owner. Never summarize a failure as only \u201cvalidation failed\u201d, \u201creview failed\u201d, or \u201cCI failed\u201d."
            }
          ]
        },
        {
          "path": "references/task-delivery-gates.md",
          "installed_sha256": "fcb02322d5f440f3e59a24fa03c4ae74370e469aa6a35a42d94db8818448cca0",
          "lines": 109,
          "remote_sha256": "2cbb0a305fccd2894046019e0ed9c7bb039203a0b1396b09888df031d8107103",
          "matches_remote": false,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/corp-v1-channel-delivery-aerosim/src/commit/0c0e68bde3189fd96415af6035200cf7fcb9c877/references/task-delivery-gates.md",
          "excerpts": [
            {
              "line": 1,
              "text": "# AeroSim Task Delivery Gates"
            },
            {
              "line": 3,
              "text": "This file contains the complete ordered procedure for delivering one AeroSim Task. Load it before selecting or continuing a Task."
            },
            {
              "line": 5,
              "text": "## End-to-End Task Delivery Gates"
            },
            {
              "line": 7,
              "text": "These gates are mandatory and ordered. Each `GATE_###_NAME` code is the stable identifier used in communication and evidence. Hermes must not skip ahead, combine evidence from different revisions, or start the next Task before `GATE_005_TASK_POST_MERGE` passes. The three repositories used by this process are:"
            },
            {
              "line": 13,
              "text": "All local work and evidence for a Task is isolated under `corp-v1-aerosim-delivery/workspace/<TASK_ID>/`. Start with the required repository clones. Create Task-local `review/`, `cache/`, `tmp/`, `logs/`, or `artifacts/` only when a concrete activity for that Task needs them. Reusable installed toolchains and technical dependencies belong in `corp-v1-aerosim-delivery/cache/` so later Tasks can reuse verified versions. Channel-provided input files belong in `corp-v1-aerosim-delivery/uploads/`."
            },
            {
              "line": 15,
              "text": "The Hermes main session owns Task selection, execution-mode choice, dependency control, final evidence verification, and communication with RootAtSkic. Delivery may run in main-runtime mode or use one isolated leaf worker per Task under `single-subagent-task-delivery.md`. RootAtSkic grants standing authorization to drain independently eligible current-release Focus work concurrently and to recover `IN_PROGRESS` Tasks that have no live executor; no repeated pickup instruction is required. Keep one worker and exclusive workspace per Task, preserve dependency order, serialize documentation merges, and converge overlapping application boundaries before later review. When several blocked Tasks need human input, discuss them with RootAtSkic one Task at a time without pausing independent workers. Never use main runtime and a worker as concurrent implementers of the same Task. Local repository tools execute deterministic commands. Gitea Actions runs remote CI. Gondor/GitOps is used only by an approved deployment or Release process. Cron never owns or continues a Task."
            },
            {
              "line": 17,
              "text": "Verification is part of each stage, not a separate end checklist. Before leaving a stage, pass its **Exit gate** and retain the evidence named there. Every Delivery progress, blocker, CI, merge, handoff, and completion update must name the active gate code and title. When a gate passes, report the passed gate code and the next gate code. At every stage, use only AeroSim's seven project channels, load the references and engineering skills required for that action, route contradictions instead of guessing, and report failures with the exact command, file, test, or CI job."
            },
            {
              "line": 19,
              "text": "### Mandatory gate-failure recovery"
            },
            {
              "line": 21,
              "text": "If any check, command, test, review, CI job, merge precondition, integration check, publication check, or Exit gate fails, the Task remains in that same gate. Hermes must diagnose the root cause, implement the smallest safe forward fix within the approved Task scope, rerun every affected check, and repeat this repair-and-verification loop until the complete gate passes. A repairable failure is work to fix; it is not a stopping point, a reason to abandon the Task, or permission to start another Task."
            },
            {
              "line": 23,
              "text": "Hermes may pause the repair loop only when progress requires an unavailable external dependency or credential, a human decision or permission, a change outside the approved requirements or Architecture, or an unsafe, destructive, or otherwise unauthorized action. In that case, keep the Task's lifecycle status unchanged, record `BLOCKED` separately with the exact evidence, responsible owner, and next action, and resume the same gate as soon as the blocker is removed. Never bypass, weaken, suppress, or relabel a failed gate to make it pass."
            },
            {
              "line": 25,
              "text": "### Stage 1 \u2014 `GATE_001_PREP` \u2014 Pick, Start, and Open the Task PR"
            },
            {
              "line": 27,
              "text": "**Repository and branch:** First read `corp-v1-aerosim/corp-v1-aerosim-documentation` remote `test` without creating a branch. After one eligible Task is identified, create `corp-v1-aerosim-delivery/workspace/<TASK_ID>/`. Create only its initial fresh repository clones: clone the application repository into `application/` and documentation repository into `documentation/`. Do not create optional support folders during preparation. In `application/`, create branch `<release>/<TASK_ID>-<short-explanation>-<attempt>` from application `test`, with a PR whose title exactly equals the branch name and whose base is `test`. In `documentation/`, create branch `<release>/<TASK_ID>-start-delivery-<attempt>` into documentation `test`, with the same title/base rules."
            },
            {
              "line": 29,
              "text": "**Runtime:** In main-runtime mode, the Hermes main session performs this stage. In single-subagent mode, the one assigned worker performs the stage from the frozen dispatch packet and returns exact evidence for main-session readback. The actor verifies eligibility, creates and reads back the application branch and PR without editing repository content, then creates the documentation branch and PR, updates the Task file, runs documentation checks, and merges the documentation PR only after its checks pass. Gitea Actions runs remote checks. Do not use cron, a second worker, or concurrent main-session implementation."
            },
            {
              "line": 31,
              "text": "**Required work:** Verify that each selected Task file says `READY_FOR_DELIVERY` or represents a recoverable `IN_PROGRESS` checkpoint, the Board places it in the current wave, dependencies have reached the states required by the Task, and required Architecture, requirements, target release, and UI/UX inputs exist. Verify parent Feature implementation approval, or for release correction work accept an exact human-authored Task `READY_FOR_DELIVERY` event that explicitly approves the current release wave and prohibits a second admission gate; historical parent portfolio status must not override that exact wave approval. Inspect only AeroSim's seven project channels. Select each independently eligible Task and assign one exclusive worker/workspace. Before mutation, create or recover its Task workspace and application/documentation clones. For a new claim, fetch application `test`; create the exact local application branch; use an empty kickoff commit if Gitea requires a difference; push; open the application PR; and verify that local branch, remote branch, PR head branch, local head, remote head, PR title, and `test` base match. Then use that Task's documentation clone to add the hash-linked `READY_FOR_DELIVERY` to `IN_PROGRESS` event with actor, time, Task, wave, application branch/PR evidence, previous hash, and new hash, and merge that documentation change. Serialize documentation merges and refresh/revalidate against latest `test` after every Task state change."
            },
            {
              "line": 33,
              "text": "**Exit gate:** The Task directory exists under `corp-v1-aerosim-delivery/workspace/<TASK_ID>/`; its required `application/` and `documentation/` clones exist and have the expected remotes; no optional support folder was created without a concrete Task need; no Task checkout was created elsewhere; the application branch and PR are visible and all identities match; the exact branch name and clickable PR URL have been written back to the matching AeroSim Delivery card; the documentation PR is merged; and the Task file on remote documentation `test` says `IN_PROGRESS`. Only then may source, test, configuration, or application documentation files be edited. Report the selected Task, wave, satisfied dependencies, application PR, Task workspace, and merged status change. If eligibility evidence is missing or contradictory, do not select or mutate the Task and do not create the application branch; stop and name the responsible owner."
            },
            {
              "line": 35,
              "text": "### Stage 2 \u2014 `GATE_002_TASK_PLAN` \u2014 Plan the Task"
            },
            {
              "line": 39,
              "text": "**Runtime:** The active executor\u2014Hermes main session in main-runtime mode or the one assigned worker in single-subagent mode\u2014inspects documentation and application code inside the Task workspace. Do not use cron, a second worker, or concurrent implementation."
            },
            {
              "line": 43,
              "text": "**Exit gate:** Every criterion has a checkable proof. A contradiction stops implementation and goes to the responsible Scope, Architecture, UI/UX, Kanban, or Release owner."
            },
            {
              "line": 45,
              "text": "### Stage 3 \u2014 `GATE_003_TASK_IMPLEMENTATION` \u2014 Implement and Review the Task"
            },
            {
              "line": 47,
              "text": "**Repository and branch:** Use `corp-v1-aerosim-delivery/workspace/<TASK_ID>/application/` on the existing application Task branch and PR. Application source, unit tests, configuration, application-documentation changes, and builds stay in that clone and branch. Every non-unit suite stays in the UAT-owned E2E repository. If a separate review checkout or output is needed, create and use `review/` under the same Task directory."
            },
            {
              "line": 49,
              "text": "**Runtime:** The active executor runs the implementation and exact-head review loops with local repository tools inside the selected Task directory. In single-subagent mode, the worker performs and records this gate, while the main session later verifies the exact-head review packet and remote effects before reporting. Create and use optional `review/`, `cache/`, `tmp/`, `logs/`, or `artifacts/` folders only when the current activity needs them. Repeat environment and real-working-directory preflight in every foreground shell, review checkout, background process, review command, and CI job. Do not use a second implementation/review worker, a human-review dependency, or cron to satisfy this gate."
            },
            {
              "line": 53,
              "text": "1. Resolve the real working directory and verify it is inside `corp-v1-aerosim-delivery/workspace/<TASK_ID>/`; then verify branch and PR identity, repository state, required tools and committed versions, and required environment-variable names without printing values. For Node/pnpm work, check the resolved `node` and `pnpm` commands and versions. Delivery unit tests use process-local fakes and must not start runtime services; UAT owns resource-dependent testing."
            },
            {
              "line": 54,
              "text": "2. For a defect, write or identify an exact named unit test and show that it fails for the expected product reason before repair. For new behavior, add a failing unit test first when practical and request the required non-unit coverage from UAT. Syntax, tooling, environment, or credentials errors do not count as product evidence."
            },
            {
              "line": 55,
              "text": "3. Implement only the approved Task, one behavior slice at a time. Keep the diff small, cover normal, edge, and failure behavior, avoid unrelated refactors, and repair forward without reverting valid work."
            },
            {
              "line": 56,
              "text": "4. Run all applicable focused and full unit tests, lint, format checks, typecheck or compile, production build, production dependency audit, static checks, generated-artifact checks, `git diff --check`, and repository-state checks. Do not run integration, browser, security-boundary, performance, container-runtime, Helm-rendering, or acceptance suites in Delivery; request and consume those results from UAT."
            },
            {
              "line": 58,
              "text": "6. The active executor reviews that exact PR head for logic, security, acceptance coverage, production-path use, and test gaps. Tests alone are not the review. The executor must inspect the complete diff and relevant production paths, record a pass/fail verdict for the exact revision, and remain accountable for that verdict. In single-subagent mode, the main session must independently verify the recorded exact-head verdict and remote identities before reporting the gate. If review finds a problem, remain in `GATE_003_TASK_IMPLEMENTATION`: repair it on the same Task branch, rerun affected local checks, push the new PR head, and review that exact head again."
            },
            {
              "line": 60,
              "text": "**Exit gate:** Every local command and review used the selected Task workspace, and every optional support folder had a concrete Task need; every acceptance criterion has implementation and test evidence; every required local check passes; the local checked revision, remote branch head, and PR head are identical; and that exact PR head has a passing review with no unresolved logic, security, acceptance, or material test finding. Any later file change keeps or returns the Task to `GATE_003_TASK_IMPLEMENTATION` until affected checks, push/readback, and review pass again. Any failure report names the exact command, file, test, or review finding; observed and expected behavior; failure class; mutation state; and repair or owner."
            },
            {
              "line": 62,
              "text": "### Stage 4 \u2014 `GATE_004_TASK_PR_CI` \u2014 Verify and Merge the Task PR"
            },
            {
              "line": 66,
              "text": "**Runtime:** Gitea Actions runs PR CI. The active executor verifies CI and PR state, confirms the exact-head review still covers the PR head, performs the authorized Gitea merge, and reads back the merge result. In single-subagent mode, the main session independently verifies those side effects before reporting. No second worker reviews or merges."
            },
            {
              "line": 70,
              "text": "1. Verify that every required Delivery CI job\u2014unit tests, lint, typecheck, build, static security checks, and artifact publication validation\u2014passes on the exact PR head reviewed in `GATE_003_TASK_IMPLEMENTATION` and within one complete successful attempt. Never combine passing jobs from different revisions or attempts."
            },
            {
              "line": 71,
              "text": "2. Immediately before merge, re-read the PR and current application `test`. Confirm that the PR remains open, its head is unchanged, its base is `test`, its branch equals the reviewed PR head, required CI is green, review still matches, the PR is mergeable, and base drift is harmless."
            },
            {
              "line": 72,
              "text": "3. If conflict resolution, rebase, or any file/tree change is required, return to `GATE_003_TASK_IMPLEMENTATION`; repeat affected local checks, push/readback, and review for the new PR head; then restart `GATE_004_TASK_PR_CI` for that exact head."
            },
            {
              "line": 75,
              "text": "**Exit gate:** All required PR CI succeeds on the exact reviewed PR head; the final pre-merge check refers to that same head; the PR is merged; and the exact application `test` integration revision is known. The Task remains `IN_PROGRESS`; merge alone does not mean `TO_BE_RELEASED` or `DONE`."
            },
            {
              "line": 77,
              "text": "### Stage 5 \u2014 `GATE_005_TASK_POST_MERGE` \u2014 Verify Integration and Hand Off the Task"
            },
            {
              "line": 79,
              "text": "**Repository and branch:** Use the selected Task's `application/` clone to validate `corp-v1-aerosim/corp-v1-aerosim` `test` at the exact integration revision produced by `GATE_004_TASK_PR_CI`. Then use the same Task's `documentation/` clone to create branch `<release>/<TASK_ID>-release-handoff-<attempt>` into documentation `test`, with a PR title exactly equal to the branch name and base `test`."
            },
            {
              "line": 81,
              "text": "**Runtime:** Gitea Actions runs application integration, build, publication, and documentation checks. The active executor verifies those results, creates and validates the documentation handoff, applies the review rules from `GATE_003_TASK_IMPLEMENTATION` to the exact documentation PR head, and performs the authorized documentation merge. In single-subagent mode, the main session independently verifies the merged canonical Task state, review identity, CI/publication evidence, cache boundary, and process cleanup before reporting delivery. Gondor/GitOps is used only if the Task explicitly includes an approved deployment."
            },
            {
              "line": 85,
              "text": "1. Verify every required application unit-test/build job, container build, immutable image publication, registry digest readback, and artifact check on the exact application `test` integration revision. PR-head CI cannot replace post-merge integration evidence."
            },
            {
              "line": 86,
              "text": "2. For every Task that changes application code, runtime behavior, images, charts, runtime configuration, or deployable assets, deploy that exact integrated artifact set to the GitOps-managed development environment as a mandatory Delivery step:"
            },
            {
              "line": 87,
              "text": "   - update the project-owned template `corp-v1-aerosim/gondor-v1-tmpl-aerosim` and render `corp-v1-aerosim/gondor-v1-aerosim-dev` through the repositories' authoritative workflow;"
            },
            {
              "line": 88,
              "text": "   - create visible review branches and PRs, validate and review their exact heads, require exact-head CI, and merge only under the existing Task authorization;"
            },
            {
              "line": 90,
              "text": "   - verify relevant pods are Ready and their runtime `imageID` values equal the exact immutable digests published from the application integration revision;"
            },
            {
              "line": 91,
              "text": "   - verify runtime source/chart annotations, configured chart version, and bounded development endpoint/readback checks match the exact deployed tuple; and"
            },
            {
              "line": 92,
              "text": "   - never substitute direct `kubectl` or Helm mutation for the reviewed GitOps path."
            },
            {
              "line": 93,
              "text": "3. A Task may record development deployment as `not_applicable` only when it is documentation-only, planning-only, test-migration, or internal tooling work with no deployable runtime effect. The Task evidence must state and verify that reason. Task wording such as \u201cif deployment is required\u201d never makes deployment optional for a runtime-affecting change."
            },
            {
              "line": 94,
              "text": "4. Hand the immutable development tuple to `corp-v1-aerosim-uat` for all applicable non-unit verification. Delivery must not execute UAT or issue its verdict."
            },
            {
              "line": 95,
              "text": "5. Add the hash-linked `IN_PROGRESS` to `TO_BE_RELEASED` event with the merged application PR, integration revision, successful integration CI, immutable publication evidence, development GitOps template/rendered PRs and revisions, Argo CD and pod-image readback, endpoint verification, UAT handoff evidence, and acceptance evidence."
            },
            {
              "line": 96,
              "text": "6. Update Ways of Working or Engineering documentation only when this Task changed how the project actually works."
            },
            {
              "line": 99,
              "text": "**Exit gate:** Every required post-merge unit-test, build, and publication job succeeds on the exact application `test` revision. For every runtime-affecting Task, the exact immutable artifacts are deployed through reviewed development GitOps, Argo CD is verified at the expected revisions as `Synced / Healthy`, relevant pods are Ready on the exact image IDs, runtime annotations and bounded endpoint/readback checks match, and the frozen tuple is handed to UAT. For a genuinely non-runtime Task, a verified `not_applicable` reason replaces those deployment fields. The documentation PR is merged; the Task file on remote documentation `test` says `TO_BE_RELEASED`; and a separate release-handoff message has been posted inside Discord channel `corp-v1-aerosim-releases` (`1537388451501056081`) and read back by message ID. A channel mention in Delivery does not satisfy this gate. The Releases-channel post must identify the Task, exact state, passed gate, and linked application, development GitOps, and documentation PR evidence, and state that release work can continue there. Only then may Delivery call the Task delivered. The post does not authorize production release execution, deployment, or the `DONE` transition."
            },
            {
              "line": 101,
              "text": "### Stage 6 \u2014 `GATE_006_NEXT_TASK` \u2014 Pick the Next Task"
            },
            {
              "line": 103,
              "text": "**Repository and branch:** Read refreshed documentation `test` and the Board from the completed Task's `documentation/` clone. Do not reuse that Task directory for another Task and create no new branch until the next Task passes `GATE_001_PREP`, which creates its own workspace and fresh clones."
            },
            {
              "line": 105,
              "text": "**Runtime:** Hermes main session selects the next Task. Do not use cron or a subagent for selection."
            },
            {
              "line": 109,
              "text": "**Exit gate:** Report the next selected Task and return to `GATE_001_PREP`. Do not wait for `DONE`; release execution and `DONE` are owned by `corp-v1-channel-releases-aerosim`."
            }
          ]
        }
      ],
      "remote_branch": "test",
      "remote_commit": "0c0e68bde3189fd96415af6035200cf7fcb9c877",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/productivity/corp-v1-channel-delivery-aerosim/SKILL.md"
    },
    {
      "skill": "corp-v1-main",
      "repository": "home-v1-skills-code-agent/corp-v1-main",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "20ed72f5b625ca4f1a05f83fbf39eb7c57646d7f4e75e7c86bfccd9563b309c2",
          "lines": 746,
          "remote_sha256": "20ed72f5b625ca4f1a05f83fbf39eb7c57646d7f4e75e7c86bfccd9563b309c2",
          "matches_remote": true,
          "remote_url": "https://gitea.lego-cloud.eu/home-v1-skills-code-agent/corp-v1-main/src/commit/23fc1ba2b68514be60c5f68b8dbf1f4f7ceb1813/SKILL.md",
          "excerpts": [
            {
              "line": 13,
              "text": "# Corp v1 Board Main Governance"
            },
            {
              "line": 15,
              "text": "For the reusable non-Discord execution checklist covering organization-scoped teams, history-preserving skill adoption, and remote artifact verification, read [`references/kickoff-execution-verification.md`](references/kickoff-execution-verification.md). For project-local authority delegations that must survive future central-reference adoption, read and follow [`references/project-authorization-overrides.md`](references/project-authorization-overrides.md). For post-kickoff Board or project-team changes, identity resolution, effective Discord permission verification, partial Gitea-access handling, record synchronization, and completion evidence, read [`references/membership-change-verification.md`](references/membership-change-verification.md). Discord operations are owned by the independently maintained `home-v1-discord` skill."
            },
            {
              "line": 17,
              "text": "## Overview"
            },
            {
              "line": 27,
              "text": "Use `home-v1-discord` for Discord discovery, provisioning, permissions, messages, pins, scheduled delivery, REST fallback, teardown, and verification. This skill owns the Board policy, registry decisions, portal publication requirements, central channel-reference change flow, and project category topology. The guild-level `corp-v1` category is reserved for Board and shared governance channels. Every project has one dedicated top-level category named `corp-v1-<code>` containing that project's exact eight lifecycle channels. At kickoff, copy and adopt the seven mandatory project engineering skills plus one central channel-reference skill for each of the eight project channels into the project's skills/code-agent organization. Reference the global `diagrams-drawio` and `corp-v1-glossary` skills by default rather than copying them. After publication, those project-scoped skills are primary across the project's Discord channels and repositories."
            },
            {
              "line": 29,
              "text": "## Board Authority"
            },
            {
              "line": 31,
              "text": "Only these Discord identities are Corp v1 Board members:"
            },
            {
              "line": 35,
              "text": "| `<@1518725627845283888>` | RootAtSkic (`root_at_skic`) |"
            },
            {
              "line": 43,
              "text": "Only a Board member listed above may:"
            },
            {
              "line": 51,
              "text": "The original seven-channel operating model, including `ui-ux` immediately after `architecture`, was approved by RootAtSkic (`discord:1518725627845283888`) in Discord message `1541719530609573890`. This decision establishes Penpot as the current platform for project frontend design work; changing the design platform later requires a reviewed update but does not require renaming the channel."
            },
            {
              "line": 53,
              "text": "The eight-channel operating model was approved by RootAtSkic (`discord:1518725627845283888`) in Discord message `1552966246591365123` on 2026-09-25. It adds mandatory `uat` between `delivery` and `releases` for every active and future Corp v1 project and requires the central `corp-v1-channel-uat` reference plus one adopted `corp-v1-channel-uat-<code>` skill per project. This supersedes the seven-channel topology while preserving the earlier order and Penpot decision."
            },
            {
              "line": 55,
              "text": "The dedicated-project-category model was approved by RootAtSkic (`discord:1518725627845283888`) in Discord message `1542625777495965836` on 2026-08-27. The guild-level `corp-v1` category retains Board and shared governance channels. Every active and future project uses one dedicated top-level Discord category named `corp-v1-<code>`; its lifecycle channel names and immutable IDs remain unchanged when existing channels are moved."
            },
            {
              "line": 57,
              "text": "The shared Corp v1 system-login credential policy was approved by RootAtSkic (`discord:1518725627845283888`) in Discord message `1542636538305712199` on 2026-08-27. When an authorized project task requires SSO login to a Corp v1 system, Hermes may use the Bitwarden-injected runtime secrets named `HL_V1_SSO_EMAIL` and `HL_V1_SSO_PASSWORD` under the credential-safety rules below. The approval records secret names and availability only; no credential value or Bitwarden object identifier may be copied into skills, repositories, Discord, logs, screenshots, or reports."
            },
            {
              "line": 59,
              "text": "The governed bootstrap-autonomy exception was approved by RootAtSkic (`discord:1518725627845283888`) in Discord message `1542278257833939087` on 2026-08-26. A project may run one 30-minute General-channel bootstrap job only when its approved kickoff proposal contains a preserved project-local authorization override, exact seven-channel boundary, explicit autonomous actions, a first-human-message stop condition, and retained safety gates. This historical decision is not widened by the later eight-channel topology; an existing override remains limited to its recorded seven channels unless separately reauthorized. New or explicitly reauthorized overrides use the current eight-channel boundary. This exception does not enable scheduler jobs for projects lacking that exact approval."
            },
            {
              "line": 61,
              "text": "Jenny (`discord:1540035254746882048`) was added to the Corp v1 Board and every active Corp v1 project by RootAtSkic (`discord:1518725627845283888`) in Discord message `1540045796152709180` on 2026-08-20. Discord access is verified; Gitea access remains pending a verified account mapping."
            },
            {
              "line": 63,
              "text": "Piotras (`discord:501954115690168321`) was added to the Corp v1 Board and every active Corp v1 project by RootAtSkic (`discord:1518725627845283888`) in Discord message `1541770529600577577` on 2026-08-25. Discord access is verified. Gitea account `piotras` (user ID `85`, `piotras.cimmperman@ibm.com`) is verified in the AeroSim project and skills teams and both Maze Next Gen organization teams, covering all repositories assigned to those teams."
            },
            {
              "line": 65,
              "text": "## When to Use"
            },
            {
              "line": 78,
              "text": "For project work after kickoff, load the relevant project-scoped skill named `<source-skill>-<code>` from `corp-v1-<code>-skills-code-agent`. Use the central source only to create, compare, or deliberately refresh the project copy. Use this `corp-v1-main` skill for Board decisions and governance-portal changes."
            },
            {
              "line": 80,
              "text": "## Non-Negotiable Rules"
            },
            {
              "line": 82,
              "text": "1. Obtain explicit approval from a listed Board member before provisioning a kickoff or executing a closure/sunset."
            },
            {
              "line": 83,
              "text": "2. Project code is lowercase, unique, recognizable, stable, and at most seven characters."
            },
            {
              "line": 85,
              "text": "4. Copy, adopt, rename, and publish all seven mandatory engineering skills and all eight mandatory channel skills; retaining only central skill references is insufficient. Use global `diagrams-drawio` unless the team explicitly approves a project copy."
            },
            {
              "line": 87,
              "text": "6. Never request, receive, read, print, or store the Harbor robot secret value; validate only that its secret name exists."
            },
            {
              "line": 88,
              "text": "7. Create one dedicated top-level Discord category named `corp-v1-<code>` and place exactly eight project channels inside it. Reserve the guild-level `corp-v1` category for Board and shared governance channels."
            },
            {
              "line": 92,
              "text": "11. Verify real remote state, Actions, and deployed artifacts before reporting completion."
            },
            {
              "line": 93,
              "text": "12. Use the project-scoped skills as primary instructions across all project Discord channels and repositories."
            },
            {
              "line": 95,
              "text": "14. Only a listed Board member may approve changes to `corp-v1-main`."
            },
            {
              "line": 96,
              "text": "15. Obtain Board approval before publishing a central `corp-v1-channel-*` reference-skill change, then ask each affected project channel to approve and perform semantic synchronization of its adopted `corp-v1-channel-*-<code>` skill."
            },
            {
              "line": 97,
              "text": "16. The default remains no Corp v1 project scheduler. Create one General-channel bootstrap job only when a Board-approved project-local override defines its cadence, channel boundary, autonomous scope, safety exclusions, and automatic stop condition. Remove or pause it immediately when that condition is met."
            },
            {
              "line": 98,
              "text": "17. For an authorized Corp v1 system login, use only the Bitwarden-injected runtime secrets `HL_V1_SSO_EMAIL` and `HL_V1_SSO_PASSWORD`; never reveal, persist, copy, or ask a human to paste their values."
            },
            {
              "line": 100,
              "text": "## Required Kickoff Metadata"
            },
            {
              "line": 102,
              "text": "Collect only missing information. Search the Board portal repository and relevant Discord history before asking the user to repeat known facts."
            },
            {
              "line": 107,
              "text": "| Project code | Unique lowercase code, maximum seven characters |"
            },
            {
              "line": 124,
              "text": "## Naming Model"
            },
            {
              "line": 129,
              "text": "Discord governance category: corp-v1 (Board and shared governance only)"
            },
            {
              "line": 132,
              "text": "                             corp-v1-<code>-scope"
            },
            {
              "line": 145,
              "text": "Engineering skill repositories: <source-skill>-<code> (seven repositories)"
            },
            {
              "line": 146,
              "text": "Channel skill repositories:     corp-v1-channel-<channel>-<code> (eight repositories)"
            },
            {
              "line": 162,
              "text": "## Approval Proposal"
            },
            {
              "line": 166,
              "text": "```markdown"
            },
            {
              "line": 167,
              "text": "## Corp v1 kickoff proposal"
            },
            {
              "line": 177,
              "text": "### Discord"
            },
            {
              "line": 178,
              "text": "- Governance category: `corp-v1` (Board and shared governance only)"
            },
            {
              "line": 182,
              "text": "  - corp-v1-<code>-scope"
            },
            {
              "line": 190,
              "text": "### Gitea"
            },
            {
              "line": 197,
              "text": "- Project engineering skills: seven repositories named <source-skill>-<code>"
            },
            {
              "line": 198,
              "text": "- Project channel skills: eight repositories named corp-v1-channel-<channel>-<code>"
            },
            {
              "line": 201,
              "text": "### Project automation"
            },
            {
              "line": 202,
              "text": "- Default: no cronjobs or scheduler jobs for Corp v1 project channels"
            },
            {
              "line": 203,
              "text": "- Approved exception: one 30-minute General-channel bootstrap job when the proposal includes a durable project-local override, exact eight-channel scope, autonomous action boundary, and first-human-message stop condition"
            },
            {
              "line": 204,
              "text": "- Existing projects retain their recorded policy until they receive separate project authorization"
            },
            {
              "line": 207,
              "text": "Explicit approval from a listed Board member authorizes only the listed scope. Present material changes for renewed Board approval."
            },
            {
              "line": 209,
              "text": "## Kickoff Execution Order"
            },
            {
              "line": 211,
              "text": "Use an idempotent discover-before-create workflow:"
            },
            {
              "line": 219,
              "text": "7. Create the dedicated `corp-v1-<code>` Discord category with the approved permission model, then create the eight lifecycle channels inside it in the required order."
            },
            {
              "line": 220,
              "text": "8. Verify the executing Hermes identity can view history, post, manage messages where needed, and operate in all eight project channels."
            },
            {
              "line": 222,
              "text": "10. Copy, adopt, rename, publish, and verify all seven mandatory engineering skills and the eight mandatory channel skills in the skills/code-agent organization; verify the global `diagrams-drawio` reference."
            },
            {
              "line": 223,
              "text": "11. Verify scheduler state against the approved project policy: absence by default, or exactly one approved General bootstrap job with the required stop condition and delivery target."
            },
            {
              "line": 226,
              "text": "14. Validate only that the required secret name exists; never access or ask for its value."
            },
            {
              "line": 230,
              "text": "18. Scaffold, validate, commit, and push all three repositories using the relevant project-scoped skills and the adopted UAT channel contract."
            },
            {
              "line": 238,
              "text": "## Discord Provisioning"
            },
            {
              "line": 240,
              "text": "`home-v1-discord` is the authoritative operational skill for all Discord mechanics. Follow it for tool selection, authenticated REST fallback, identity resolution, idempotent channel mutations, effective permissions, messages, pins, deletion, and remote readback. The rules below are Corp v1 policy and scope only."
            },
            {
              "line": 254,
              "text": "Copy only the approved permission model needed for the project; do not weaken unrelated guild or governance access. Create exactly these text channels inside the dedicated project category:"
            },
            {
              "line": 259,
              "text": "| `corp-v1-<code>-scope` | Roadmap, Ideas, Epics, Features, value, acceptance outcomes, and human approval for solution |"
            },
            {
              "line": 267,
              "text": "Place `corp-v1-<code>-ui-ux` immediately below `corp-v1-<code>-architecture` in the dedicated `corp-v1-<code>` category. Keep the complete channel order `general`, `scope`, `architecture`, `ui-ux`, `kanban`, `delivery`, `uat`, `releases`."
            },
            {
              "line": 269,
              "text": "Do not create extra project categories, voice channels, decisions channels, support channels, or other channels during kickoff. The dedicated `corp-v1-<code>` category is the only project category."
            },
            {
              "line": 280,
              "text": "Verify dedicated-category placement, exact channel names, purpose messages, and the pinned introduction."
            },
            {
              "line": 282,
              "text": "### Existing-project category migration"
            },
            {
              "line": 284,
              "text": "When adopting this policy for an existing project, preserve every channel ID, message history, topic, permission overwrite, pin, webhook, and scheduled-delivery target. Discover the complete guild inventory first; create or reuse the exact `corp-v1-<code>` category; copy the currently approved project permission model without weakening access; move only the project's exact lifecycle channel IDs; create the missing `corp-v1-<code>-uat` channel when absent; apply the required order; and read back every channel's `parent_id`, position, name, and type. Keep `corp-v1-board` and shared governance channels in the guild-level `corp-v1` category. The Board decision in Discord message `1552966246591365123` authorizes the eight-channel policy, but each active-project migration must still preserve project permissions, install and adopt `corp-v1-channel-uat-<code>`, update channel routing and records, and verify the resulting remote state. A central-skill update alone is not evidence that an existing project received its UAT channel."
            },
            {
              "line": 286,
              "text": "### Hermes access and Corp project automation policy"
            },
            {
              "line": 290,
              "text": "Corp v1 projects and their Discord channels do not use scheduler jobs by default. The only permitted exception is one Board-approved General-channel bootstrap job recorded in the kickoff proposal and a preserved `CORP_V1_PROJECT_AUTHORIZATION_OVERRIDE` block. It must run at the approved cadence, inspect only the exact channel IDs recorded in that override, use the project-adopted skills, deliver only to General, and check for human-authored messages in that recorded channel set before every mutation. Historical seven-channel overrides remain seven-channel; UAT is added to a bootstrap boundary only through a new or amended explicit authorization. New eight-channel projects record all eight IDs. The first qualifying human message ends bootstrap autonomy: the run must perform no further autonomous mutation, report the stop evidence, and pause or remove the job. Spending, secret values, destructive operations, vendor-account changes, legal commitments, and production-payment go-live remain outside the exception. Existing projects without this exact approval retain scheduler absence."
            },
            {
              "line": 292,
              "text": "## Corp v1 System SSO Credential Use"
            },
            {
              "line": 303,
              "text": "1. Confirm the requested hostname and login purpose belong to the authorized task before filling any credential field. Secret availability is capability, not permission to browse or mutate unrelated systems."
            },
            {
              "line": 304,
              "text": "2. Check only whether both runtime variables are present. Never print, echo, serialize, hash, measure, inspect, return, or log their values."
            },
            {
              "line": 305,
              "text": "3. Use credentials only in a trusted login form or provider-supported secure runtime interface. Never place them in a command line, URL, repository, file, patch, prompt, Discord message, screenshot, browser console, test fixture, CI output, or generated artifact."
            },
            {
              "line": 306,
              "text": "4. Never ask a human to paste either value into Discord or chat. If either variable is unavailable, report only the missing secret name and request Bitwarden/gateway injection or restart; do not create a local fallback secret."
            },
            {
              "line": 307,
              "text": "5. Verify the destination origin before submission, reject lookalike or unexpected identity-provider hosts, and do not bypass TLS, MFA, access policy, or account protections."
            },
            {
              "line": 309,
              "text": "7. Do not expose the SSO identity in ordinary reports. Record only the target system, authorized purpose, login success/failure, and non-sensitive verification evidence. Redact credential-bearing errors and avoid screenshots while credential fields or sensitive account details are visible."
            },
            {
              "line": 312,
              "text": "## Gitea Organizations and Team"
            },
            {
              "line": 326,
              "text": "Do not place application or documentation repositories in the skills/code-agent organization. It is the dedicated boundary for reusable project skills, code-agent instructions, supporting assets, and future agent-oriented repositories. The seven mandatory engineering-skill repositories and eight mandatory channel-skill repositories are part of minimum kickoff; any additional repository there requires an explicitly approved purpose."
            },
            {
              "line": 335,
              "text": "- every username is verified before membership is assigned;"
            },
            {
              "line": 348,
              "text": "## Mandatory Project-Scoped Skills"
            },
            {
              "line": 356,
              "text": "At kickoff, copy and adopt these exact source repositories from their current default branches:"
            },
            {
              "line": 364,
              "text": "| `devsecops-ci-cd-gitea` | `devsecops-ci-cd-gitea-<code>` | Gitea Actions, Harbor authentication, build, validation, publication, and delivery gates |"
            },
            {
              "line": 368,
              "text": "Also copy and adopt these exact central channel-reference repositories, one mapped to each project channel:"
            },
            {
              "line": 373,
              "text": "| `corp-v1-channel-scope` | `corp-v1-channel-scope-<code>` | `corp-v1-<code>-scope` |"
            },
            {
              "line": 392,
              "text": "4. Change the skill frontmatter `name:` to exactly `<source-skill>-<code>` and update its title, description, examples, repository links, environment namespace, and project assumptions where required."
            },
            {
              "line": 393,
              "text": "5. Preserve attribution to the central source and record the source repository URL, source branch, and source commit used for adoption. Do not rename references to other skills unless matching project-scoped copies exist."
            },
            {
              "line": 395,
              "text": "7. Keep every useful source asset. Delete an asset only when it is demonstrably irrelevant and record that adoption decision."
            },
            {
              "line": 398,
              "text": "Do not mutate the fifteen copied central source repositories during project adoption. The project repositories are independent adopted copies; future central changes do not silently overwrite project decisions. Refreshing a project copy requires a deliberate comparison, reviewed adoption, validation, and recorded source commit."
            },
            {
              "line": 400,
              "text": "### Central reference-skill change rollout"
            },
            {
              "line": 402,
              "text": "A change to any central `corp-v1-channel-*` reference skill is incomplete until the Board approves the central proposal and every existing project adoption has been deliberately assessed. Treat central publication and project synchronization as one governed operation:"
            },
            {
              "line": 404,
              "text": "1. Inventory all `corp-v1-*-skills-code-agent` organizations and every matching `corp-v1-channel-<channel>-<code>` repository; do not rely only on local clones or the currently active project."
            },
            {
              "line": 405,
              "text": "2. Present the exact central reference-skill proposal and impact inventory to the Board. Do not publish until a listed Board member explicitly approves it; record the approving Discord ID and message evidence."
            },
            {
              "line": 408,
              "text": "5. Before changing each project copy, load it and extract every `CORP_V1_PROJECT_AUTHORIZATION_OVERRIDE` block described in [`references/project-authorization-overrides.md`](references/project-authorization-overrides.md). Fail closed on malformed or unexpectedly changed markers."
            },
            {
              "line": 411,
              "text": "8. Validate that every pre-update override marker and `override_id` remains present exactly once, then commit, push, independently read back every approved updated project repository, and record the Board-approved central source commit, project-channel approval evidence, and project adoption commit."
            },
            {
              "line": 412,
              "text": "9. Refresh any local/runtime project skill copy used by agents, then verify each project's scheduler state against its recorded policy: absent by default or the one exact approved bootstrap job."
            },
            {
              "line": 413,
              "text": "10. Report exact coverage: Board approval evidence, central repository and SHA, discovered project copies, preserved override IDs, project channels asked, approved and updated copies, pending/skipped copies with reasons, remote SHAs, scheduler-policy verification, and CI status or the explicit fact that CI is not configured."
            },
            {
              "line": 417,
              "text": "### Primary use across project work"
            },
            {
              "line": 419,
              "text": "After publication, references such as `development-scripts` mean the project-scoped `development-scripts-<code>` unless the task explicitly concerns the central source. Pin the seven engineering-skill repository URLs and responsibilities in all eight same-project Discord channels, pin each channel-skill repository in its mapped channel, reference global `diagrams-drawio` for diagram work, and link all fifteen project copies from all three project repositories. Code agents operating in a project channel must load its mapped channel skill plus any relevant project engineering/global skills before acting."
            },
            {
              "line": 434,
              "text": "When several domains apply, load all applicable project-scoped skills and resolve conflicts in favor of explicit Corp v1 requirements and approved project decisions."
            },
            {
              "line": 436,
              "text": "## Project Organization Harbor Actions Configuration"
            },
            {
              "line": 458,
              "text": "4. read back both variables and verify their exact names and values."
            },
            {
              "line": 460,
              "text": "These two shared infrastructure variables are the only explicit `HL_V1_*` exception in this skill. Project-specific application variables still use `CORP_V1_<PROJECT_CODE>_*`."
            },
            {
              "line": 468,
              "text": "The request must identify the exact organization and secret name, but it must **not** ask anyone to send or paste the secret value in Discord or to Hermes. Members enter the value directly in Gitea's organization Actions secrets UI."
            },
            {
              "line": 476,
              "text": "Paginate when required and verify only that an entry named `HL_V1_HARBOR_ROBOT_GITEA_ACTIONS_V1_SECRET` exists. The list response exposes metadata such as name and creation time, not the secret value. Never call the create/update secret endpoint, request the value, print it, store it, commit it, or attempt to test it by echoing it. A missing secret is a kickoff blocker for Harbor-dependent Actions; report the exact pending member action and resume after confirmation."
            },
            {
              "line": 478,
              "text": "## Documentation Repository"
            },
            {
              "line": 487,
              "text": "Use `documentation-docusaurus-<code>` as the primary implementation and authoring instruction, `devsecops-ci-cd-gitea-<code>` for Actions/Pages delivery, and global `diagrams-drawio` for documentation diagrams unless an approved project copy is recorded. The Corp lifecycle retains these required outcomes:"
            },
            {
              "line": 494,
              "text": "- Gitea Actions publication through Gondor v1 Gitea Pages."
            },
            {
              "line": 499,
              "text": "2. Scope"
            },
            {
              "line": 506,
              "text": "`Glossary` is always last and has its own dedicated sidebar beginning with `Terms`. Reusable term definitions belong there. Domain pages retain domain evidence, decisions, requirements, and procedures and link to the canonical Glossary entry rather than embedding generic term explanations."
            },
            {
              "line": 508,
              "text": "`Kanban` must appear immediately after `Architecture` and use its own sidebar ordered `Board`, then `Focus`. Board shows unfinished Epics and Features in `InIdeation`, `InBacklog`, `InProgress`, or `ToBeReleased`; terminal work is excluded. Focus shows only Epic/Feature/task hierarchy in `InBacklog`, `InProgress`, or `ToBeReleased`. A task may enter Focus `InBacklog` only through explicit human team-member approval; Delivery may execute only such admitted tasks whose Feature is also approved for implementation."
            },
            {
              "line": 510,
              "text": "Each area must have its own documentation section and dedicated sidebar. The sidebar item within each section may use the concise label `Registry` where appropriate, but top-level project documentation areas retain the names above."
            },
            {
              "line": 518,
              "text": "- links to all seven documentation areas;"
            },
            {
              "line": 525,
              "text": "## Application Repository"
            },
            {
              "line": 534,
              "text": "Use these project-scoped skills as primary implementation instructions:"
            },
            {
              "line": 538,
              "text": "- `devsecops-ci-cd-gitea-<code>` for build, validation, Harbor publication, and Gitea Actions;"
            },
            {
              "line": 543,
              "text": "The Corp lifecycle retains only these project-level decisions and outcomes:"
            },
            {
              "line": 547,
              "text": "- independently buildable applications under `applications/`;"
            },
            {
              "line": 549,
              "text": "- Docker images and Helm charts for deployable applications;"
            },
            {
              "line": 550,
              "text": "- Taskfile only, never Makefile;"
            },
            {
              "line": 552,
              "text": "- Gitea Actions validation and Harbor publication where applicable;"
            },
            {
              "line": 565,
              "text": "## E2E and UAT Repository"
            },
            {
              "line": 574,
              "text": "Use the adopted `corp-v1-channel-uat-<code>` skill as the primary repository and execution contract. The repository is a standalone pnpm/Playwright test repository; it is not part of the deployable application monorepo and contains no application implementation, Kubernetes manifests, Helm charts, Argo CD mutation, image publication, or release promotion."
            },
            {
              "line": 579,
              "text": "- explicit non-production base URL and runtime-only authentication;"
            },
            {
              "line": 582,
              "text": "- smoke journeys for an explicit PR event or an exact commit represented by a Gitea pull-request head ref; retained closed-PR refs may conservatively cause extra smoke rather than exposing a broad API token to candidate-controlled code;"
            },
            {
              "line": 583,
              "text": "- one queued `UAT / Integrated Regression` workflow for each exact commit merged into `test`;"
            },
            {
              "line": 584,
              "text": "- structured JUnit/JSON and failure traces/screenshots, with video only under the project privacy contract;"
            },
            {
              "line": 585,
              "text": "- a separate live deployment-tuple gate before any release-facing `UAT_PASSED` verdict;"
            },
            {
              "line": 590,
              "text": "## Board Portal Records"
            },
            {
              "line": 602,
              "text": "- scheduler policy and verification: absence by default, or the approved bootstrap job ID, cadence, General delivery target, override ID, and stop condition;"
            },
            {
              "line": 605,
              "text": "- all fifteen project-scoped skill repository URLs, adopted names, central source branches, and source commits, including each channel-skill mapping;"
            },
            {
              "line": 608,
              "text": "- Harbor secret presence status only\u2014never its value;"
            },
            {
              "line": 618,
              "text": "## Kickoff Definition of Done"
            },
            {
              "line": 621,
              "text": "- [ ] Project code is unique and no longer than seven characters."
            },
            {
              "line": 622,
              "text": "- [ ] Dedicated `corp-v1-<code>` project category exists; all eight exact Discord channels are inside it in lifecycle order; Board/shared governance channels remain under `corp-v1`."
            },
            {
              "line": 624,
              "text": "- [ ] Hermes/code-agent access is verified in all eight project channels."
            },
            {
              "line": 625,
              "text": "- [ ] Scheduler inventory matches the approved project policy: none by default, or exactly one approved General bootstrap job with durable override and stop-condition evidence."
            },
            {
              "line": 628,
              "text": "- [ ] All seven engineering-skill repositories and eight channel-skill repositories exist on default branch `test` in the skills/code-agent organization."
            },
            {
              "line": 630,
              "text": "- [ ] Each copied `SKILL.md` has the exact suffixed `name:`, project adoption, and recorded source branch/commit."
            },
            {
              "line": 631,
              "text": "- [ ] Project skill repositories and their responsibilities are pinned in all eight same-project Discord channels and linked from all three project repositories."
            },
            {
              "line": 632,
              "text": "- [ ] Both required Harbor organization Actions variables exist with the exact approved values on `corp-v1-<code>`."
            },
            {
              "line": 638,
              "text": "- [ ] Documentation has an entry page and seven required top-level areas with dedicated sidebars; Glossary is last and begins with Terms."
            },
            {
              "line": 642,
              "text": "- [ ] Relevant project-scoped development skills pass, including Taskfile and absence of Makefile."
            },
            {
              "line": 643,
              "text": "- [ ] Variables use only `CORP_V1_<PROJECT_CODE>_*`."
            },
            {
              "line": 650,
              "text": "## Project Sunset"
            },
            {
              "line": 656,
              "text": "1. Read current Board portal records and inventory channels, repositories, deployments, automation, credentials, access, retained data, dependencies, and open obligations."
            },
            {
              "line": 657,
              "text": "2. Present the sunset scope and receive explicit approval from a listed Board member; record the approving Discord ID and message evidence."
            },
            {
              "line": 658,
              "text": "3. Record the sunset decision, reason, owner, final outcome, closure date, retained artifacts, and unresolved obligations."
            },
            {
              "line": 660,
              "text": "5. Post a closure notice in `corp-v1-<code>-general` with date and retained links."
            },
            {
              "line": 662,
              "text": "7. Verify that no cronjob or scheduler job targets the project channels; remove any discovered legacy job."
            },
            {
              "line": 665,
              "text": "   - `corp-v1-<code>-scope`"
            },
            {
              "line": 677,
              "text": "11. Archive all fifteen required project-scoped skill repositories, plus any explicitly approved project `diagrams-drawio` copy, in `corp-v1-<code>-skills-code-agent` and verify their retained source/adoption records."
            },
            {
              "line": 678,
              "text": "12. Verify all eighteen required repositories, plus any optional approved Draw.io copy, are marked archived and effectively read-only."
            },
            {
              "line": 680,
              "text": "14. Retain both dedicated Gitea organizations as historical containers unless steering explicitly approves their deletion."
            },
            {
              "line": 686,
              "text": "## Sunset Definition of Done"
            },
            {
              "line": 689,
              "text": "- [ ] Final outcome, retained artifacts, and unresolved obligations are documented."
            },
            {
              "line": 691,
              "text": "- [ ] No cronjob or scheduler job targets the project channels."
            },
            {
              "line": 692,
              "text": "- [ ] All three project repositories and all fifteen required project-skill repositories, plus any approved Draw.io copy, are archived and verified read-only."
            },
            {
              "line": 693,
              "text": "- [ ] Both Gitea organizations are retained unless their deletion was separately approved."
            },
            {
              "line": 695,
              "text": "- [ ] Required documentation and retained artifacts remain accessible."
            },
            {
              "line": 699,
              "text": "## Common Pitfalls"
            },
            {
              "line": 701,
              "text": "1. Provisioning resources before the complete proposal is approved."
            },
            {
              "line": 702,
              "text": "2. Accepting a code longer than seven characters or inconsistent across systems."
            },
            {
              "line": 703,
              "text": "3. Reusing the guild-level `corp-v1` governance category for project lifecycle channels, creating extra project categories, or placing a project channel outside its exact `corp-v1-<code>` category."
            },
            {
              "line": 707,
              "text": "7. Linking central skills without creating and adopting all seven engineering-skill and eight channel-skill project repositories."
            },
            {
              "line": 708,
              "text": "8. Renaming only the repository while leaving `name:` or internal project assumptions unchanged."
            },
            {
              "line": 710,
              "text": "10. Omitting the documentation entry page or one of the seven required areas, or placing Kanban anywhere other than immediately after Architecture."
            },
            {
              "line": 715,
              "text": "15. Using `HL_V1_*` outside the three approved shared Harbor Actions names, or using unscoped project variables."
            },
            {
              "line": 717,
              "text": "17. Claiming the Harbor secret is configured without listing organization Actions secrets and matching its exact name."
            },
            {
              "line": 718,
              "text": "18. Letting cross-channel coordination inspect channels from another project instead of only the eight explicitly approved same-project channels."
            },
            {
              "line": 723,
              "text": "23. Deleting Discord channels before preserving required history and verifying that no legacy scheduler job targets them."
            },
            {
              "line": 725,
              "text": "25. Creating, retaining, or testing a scheduler job without the exact Board-approved project override and stop condition."
            },
            {
              "line": 731,
              "text": "## Final Verification Checklist"
            },
            {
              "line": 734,
              "text": "- [ ] Names, identities, and existing resources were discovered before mutation."
            },
            {
              "line": 735,
              "text": "- [ ] Every created or changed resource was read back independently."
            },
            {
              "line": 736,
              "text": "- [ ] All fifteen project-scoped skills are correctly suffixed, adopted, mapped, primary, and remotely verified; global `diagrams-drawio` and `corp-v1-glossary` are referenced unless an explicit project-copy decision exists."
            },
            {
              "line": 739,
              "text": "- [ ] Harbor organization Actions variables match their exact approved names and values."
            },
            {
              "line": 740,
              "text": "- [ ] Harbor secret validation checked name presence only and did not access or expose its value."
            },
            {
              "line": 744,
              "text": "- [ ] Discord placement, Hermes access, names, messages, pins, and scheduler state against the project policy were verified."
            }
          ]
        }
      ],
      "remote_branch": "test",
      "remote_commit": "23fc1ba2b68514be60c5f68b8dbf1f4f7ceb1813",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/productivity/corp-v1-main/SKILL.md"
    },
    {
      "skill": "corp-v1-channel-general-aerosim",
      "repository": "corp-v1-aerosim-skills-code-agent/corp-v1-channel-general-aerosim",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "97c4d13a3ebbb0a180464fd946572a7121fc8e879827903dcdbb2282938ea2b3",
          "lines": 363,
          "remote_sha256": "7a3e32520b4865a0a3b25626557a2dc807a70dfcf00a00f3f5d269aa789a2a34",
          "matches_remote": false,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/corp-v1-channel-general-aerosim/src/commit/f97a4e8bbc1e107ddff0fac02a6730a32afed56e/SKILL.md",
          "excerpts": [
            {
              "line": 3,
              "text": "description: Coordinate AeroSim across its eight project channels."
            },
            {
              "line": 13,
              "text": "# AeroSim General Channel"
            },
            {
              "line": 15,
              "text": "<!-- AEROSIM_BLOCKER_OWNERSHIP:BEGIN -->"
            },
            {
              "line": 16,
              "text": "## Hermes owns blocker resolution"
            },
            {
              "line": 18,
              "text": "RootAtSkic's [project-wide correction](https://discord.com/channels/1518726359512387766/1537227683191197807/1554922326859063327) makes Hermes main session accountable for resolving blockers and coordinating workers, not merely reporting blocked cards or asking the human to manage the queue. Preserve this project-local rule during adoption and synchronization."
            },
            {
              "line": 20,
              "text": "On a blocked notification, inspect the existing card, dependency graph, worker evidence and current source; do not recreate the task or graph. Diagnose and resolve repairable technical, toolchain, CI, dependency and bookkeeping failures within existing authorization. Check current standing authority before requesting permission again. A compatible security patch may be a narrow gate repair; establish that from the actual dependency change and existing scope, rather than automatically treating every audit finding as a human decision. Never waive security checks or presume that all dependency upgrades are authorized."
            },
            {
              "line": 22,
              "text": "Reconcile missing records from explicit existing decisions through the owning channel workflow; never invent product approval or expand scope. Route cross-channel work to its owner and follow through: saying \u201canother channel must resolve it\u201d is not resolution. Reuse the existing worker/card, resume only after the recorded blocker is addressed, and verify execution and results; merely unblocking or queueing a card is not active work or a resolved issue. Keep independently executable work moving without duplicate workers."
            },
            {
              "line": 24,
              "text": "Escalate only a genuine missing human decision, unavailable access/capability, or action outside standing authority. Present one concrete decision at a time, with evidence, the minimal required action and a recommendation\u2014not a list of internal chores. Preserve human Scope/UI/UX/Architecture ownership, review and exact-head CI, security gates, UAT acceptance, and separately authorized production/release actions. A complaint about delay is not blanket approval to bypass them."
            },
            {
              "line": 26,
              "text": "Verification: every blocker has an accountable executor and next action; recoverable work is actually resumed; remaining escalation names the precise unsatisfied boundary; status distinguishes queued, running, verified and deployed."
            },
            {
              "line": 27,
              "text": "<!-- AEROSIM_BLOCKER_OWNERSHIP:END -->"
            },
            {
              "line": 30,
              "text": "## Standing AeroSim PR merge authority"
            },
            {
              "line": 32,
              "text": "RootAtSkic authorizes Hermes to merge PRs across all AeroSim repositories and channels without per-PR human authorization. Source: [explicit standing delegation](https://discord.com/channels/1518726359512387766/1537544173366943886/1554873863165182015)."
            },
            {
              "line": 34,
              "text": "This project-local rule supersedes older instructions in this skill, linked references, or inherited guidance requiring a human to authorize PR merging or forbidding Hermes from merging. Preserve it during skill adoption and synchronization. Complete required review and exact-head CI, freshly verify the unchanged mergeable head, merge it, and verify the integration revision and required post-merge checks. Do not block a reviewed, green PR for owner merge approval or request repeated permission."
            },
            {
              "line": 36,
              "text": "This delegates PR integration, not product/design/architecture decisions, human UAT acceptance, or independently gated release/deployment actions. If merging a GitOps or release PR itself triggers a separately gated deployment, establish that deployment authority first; do not relabel it as a PR-merge approval. Repository protections, security checks, and genuine technical blockers still apply."
            },
            {
              "line": 40,
              "text": "## Overview"
            },
            {
              "line": 42,
              "text": "## Hermes Kanban direct-link contract"
            },
            {
              "line": 44,
              "text": "Whenever a response mentions this project's Hermes Kanban board, initiative, card, or task, it must include the direct board link: [AeroSim UAT and GitOps board](https://jarvis-jr-v1.lego-cloud.eu/kanban?board=aerosim-uat-gitops). Never report only the board name, slug, card count, or wording such as \u201cvisible cards\u201d or \u201cavailable in the dashboard.\u201d If no task-specific deep link exists, link the board and include the exact task ID or title in the same statement."
            },
            {
              "line": 47,
              "text": "## Dedicated Channel Workspace"
            },
            {
              "line": 54,
              "text": "Keep new operational files for this channel inside this root rather than placing them directly in `/opt/data`. Use run-unique or task-specific subdirectories under `workspace/`; keep durable project records in the approved AeroSim repositories. Do not place another project's work in this folder. The folder boundary organizes local execution only and does not broaden this channel's authority."
            },
            {
              "line": 57,
              "text": "<!-- AEROSIM_PROJECT_OWNERSHIP:BEGIN -->"
            },
            {
              "line": 58,
              "text": "## Project-local Human Ownership \u2014 Preserve During Adoption"
            },
            {
              "line": 61,
              "text": "ownership_id: aerosim-domain-ownership-v3"
            },
            {
              "line": 64,
              "text": "primary_owners:"
            },
            {
              "line": 65,
              "text": "  scope: discord:773109855497093191"
            },
            {
              "line": 68,
              "text": "backup_owner:"
            },
            {
              "line": 71,
              "text": "scope: all-eight-project-channels"
            },
            {
              "line": 75,
              "text": "The named human is the primary accountable owner and routine decision/approval authority for that domain. Scope decisions route to `<@773109855497093191>`, UI/UX decisions route to Dovydas (`<@1543952051212656684>`), and Architecture decisions route to `<@1543940246071410706>`. RootAtSkic (`<@1518725627845283888>`) retains backup ownership across every AeroSim channel: RootAtSkic may act when a primary owner is unavailable, when work is explicitly escalated, to resolve cross-domain conflict, or to override a domain decision."
            },
            {
              "line": 77,
              "text": "No new primary owner is assigned here for General, Kanban, Delivery, or Releases; RootAtSkic retains accountable ownership for those domains. Hermes remains the operational and coordination agent: it may research, draft, implement approved work, validate, publish, merge, release, deploy, and verify only where the owning channel workflow already authorizes those actions. Hermes must not substitute itself for the named Scope, UI/UX, or Architecture owner when an exact domain decision or approval is required unless RootAtSkic explicitly delegates that exact action."
            },
            {
              "line": 79,
              "text": "RootAtSkic's 2026-09-28 decision supersedes ownership version 2 for Scope only: Jone Tamulaite - Nong left AeroSim and MartynasP (`<@773109855497093191>`) became the primary Scope owner. Preserve earlier decisions as history; route new Scope decisions to MartynasP."
            },
            {
              "line": 81,
              "text": "This decision supersedes the earlier broad Hermes project-lead delegation only where it conflicts with the three named human domain owners. It does not waive exact-item decision records, canonical evidence, accessibility review, CI, security/quality gates, external permissions, spending controls, legal commitments, or confirmation for destructive or irreversible high-impact operations. This project-local block must be preserved during central/reference skill adoption."
            },
            {
              "line": 82,
              "text": "<!-- AEROSIM_PROJECT_OWNERSHIP:END -->"
            },
            {
              "line": 86,
              "text": "This skill owns project-level awareness and coordination for a Corp v1 project's `general` channel. It monitors the complete same-project channel set\u2014`general`, `scope`, `architecture`, `ui-ux`, `kanban`, `delivery`, `uat`, and `releases`\u2014and turns verified activity into concise overall updates, issue visibility, and maintained guidance. Scope owns the authoritative Roadmap, Ideas, Epics, and Features; General may propose improvements and routes them to Scope instead of duplicating records."
            },
            {
              "line": 88,
              "text": "It is not a replacement for architecture, delivery, or release ownership. It connects those workstreams, makes their implications visible to the whole project, and routes specialized work to the corresponding channel."
            },
            {
              "line": 90,
              "text": "Load the project-adopted `documentation-docusaurus-aerosim` skill from `https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/documentation-docusaurus-aerosim` before changing documentation structure, navigation, Markdown/MDX, Docusaurus configuration, data-backed React components, or builds."
            },
            {
              "line": 92,
              "text": "Load the global `corp-v1-glossary` skill from `https://gitea.lego-cloud.eu/home-v1-skills-code-agent/corp-v1-glossary` whenever project documentation needs to define or explain a reusable term. Maintain one canonical definition in the project's final top-level **Glossary** area and link to it from the owning domain page; do not duplicate glossary-style explanations across channel documentation."
            },
            {
              "line": 94,
              "text": "## Shared Corp v1 System Login"
            },
            {
              "line": 96,
              "text": "When an authorized task requires login to a Corp v1 system being built or operated, follow the shared policy in `corp-v1-main` and use only the Bitwarden-injected runtime secrets named:"
            },
            {
              "line": 103,
              "text": "Secret availability is capability, not authorization. Verify the destination origin and task purpose before login. Never print, inspect, log, hash, serialize, paste, screenshot, or persist either value; never place a value in a command line, URL, file, repository, prompt, Discord message, browser console, test fixture, CI output, or generated artifact. Never ask a human to paste a value into chat. If a variable is unavailable, report only its missing name and request Bitwarden/gateway injection. Login does not authorize account recovery, MFA or credential changes, permission changes, billing, spending, destructive operations, or access outside the approved project task."
            },
            {
              "line": 105,
              "text": "## When to Use"
            },
            {
              "line": 110,
              "text": "- running the recurring synchronization job mapped to that channel;"
            },
            {
              "line": 112,
              "text": "- detecting cross-channel issues, ownership gaps, or conflicting direction;"
            },
            {
              "line": 118,
              "text": "## Approved Information Boundary"
            },
            {
              "line": 120,
              "text": "Read only the eight channels belonging to the same project:"
            },
            {
              "line": 124,
              "text": "corp-v1-aerosim-scope"
            },
            {
              "line": 133,
              "text": "Read enough mapped-channel history to avoid duplicate updates and only new relevant messages from the other seven channels since the previous useful synchronization. Never correlate another project's channels."
            },
            {
              "line": 135,
              "text": "## Core Responsibilities"
            },
            {
              "line": 137,
              "text": "### 1. Provide overall updates"
            },
            {
              "line": 139,
              "text": "Produce concise updates based on evidence, covering only substantive changes:"
            },
            {
              "line": 145,
              "text": "- responsible owners and next actions;"
            },
            {
              "line": 150,
              "text": "```markdown"
            },
            {
              "line": 151,
              "text": "## Project update"
            },
            {
              "line": 161,
              "text": "### 2. Propose Epics, Features, and improvements"
            },
            {
              "line": 163,
              "text": "Identify opportunities grounded in project messages, requirements, architecture, delivery evidence, user feedback, or release outcomes. Hermes is an active participant in discovery: on every synchronization iteration, add or materially improve at least one useful project artifact when safe and evidence permits. Prefer refining an existing item over creating noise or duplicates."
            },
            {
              "line": 165,
              "text": "All Epics, Features, and Ideas must live in the project documentation repository, not only in Discord. Use the uppercase project code in stable identifiers:"
            },
            {
              "line": 177,
              "text": "Epic: ID | title | problem/opportunity | intended outcome | value | status | owner | feature links | evidence"
            },
            {
              "line": 178,
              "text": "Feature: ID | epic | user/problem statement | expected value | scope | acceptance outcomes | status | owner | dependencies | evidence"
            },
            {
              "line": 183,
              "text": "- the observed problem or opportunity;"
            },
            {
              "line": 187,
              "text": "- unknowns and risks;"
            },
            {
              "line": 188,
              "text": "- proposed owner and destination channel;"
            },
            {
              "line": 191,
              "text": "Use lifecycle states `Proposed`, `Approved for Solution`, `Rejected`, or `Deferred`. Only the primary Scope owner `<@773109855497093191>` or RootAtSkic as backup may move an Epic or Feature to `Approved for Solution`. Hermes may propose and refine items, but must never approve its own proposal or imply approval from silence."
            },
            {
              "line": 193,
              "text": "## Project Documentation Ownership"
            },
            {
              "line": 197,
              "text": "### Mandatory Scope information architecture"
            },
            {
              "line": 199,
              "text": "Every project documentation site must expose **Scope** as a top-level navigation item immediately **before Architecture**. Scope must be a dedicated documentation area with its own sidebar\u2014not a category inside another area's sidebar."
            },
            {
              "line": 201,
              "text": "The Scope sidebar order and hierarchy are mandatory:"
            },
            {
              "line": 217,
              "text": "2. **Epics** is the ordered registry and hierarchy for all Epic records. Each Epic has its own page titled with its stable `<CODE>-EP-N` ID and human-readable title."
            },
            {
              "line": 219,
              "text": "4. Each **Feature** is nested under its parent Epic in the sidebar and has its own page titled with its stable `<CODE>-FT-N` ID and human-readable title."
            },
            {
              "line": 220,
              "text": "5. Each **Feature page** contains at least the Feature title, description, and a Tasks section with a table linking every architecture-owned implementation task and showing its current status. Preserve the task identifiers established by the Architecture channel; do not create duplicate General-channel task records."
            },
            {
              "line": 221,
              "text": "6. **Ideas** tracks early opportunities that are not yet accepted as Epics or Features. Each Idea records a stable local reference, title, description, evidence/source, owner when known, status, and any resulting Epic/Feature link. Promotion preserves provenance rather than deleting the Idea."
            },
            {
              "line": 225,
              "text": "```markdown"
            },
            {
              "line": 233,
              "text": "```markdown"
            },
            {
              "line": 239,
              "text": "Validate top-menu order, Scope's dedicated sidebar, hierarchy, unique IDs, parent-child relationships, and every Epic\u2192Feature and Feature\u2192Task link during documentation builds. Empty registries require useful empty-state pages; they do not justify omitting Scope, Roadmap, Epics, or Ideas."
            },
            {
              "line": 241,
              "text": "Each useful iteration must add or materially improve at least one evidence-based artifact\u2014for example a grounded proposal, better value/scope/acceptance outcomes, deduplication, status from an explicit human decision, cross-links, or a concrete unanswered product question. Never create filler merely to satisfy cadence. If evidence is insufficient, document and raise the specific blocker/question."
            },
            {
              "line": 243,
              "text": "### 3. Monitor faced issues"
            },
            {
              "line": 251,
              "text": "- missing ownership, contradictory instructions, and stale documentation."
            },
            {
              "line": 253,
              "text": "Deduplicate repeated reports, link evidence, track owner/status when known, and route technical handling to the appropriate channel. Escalate unresolved cross-channel blockers in `general` without copying every low-level log."
            },
            {
              "line": 255,
              "text": "### 4. Maintain current guidance"
            },
            {
              "line": 270,
              "text": "## Kanban Handoff"
            },
            {
              "line": 272,
              "text": "General preserves authoritative Epic/Feature identity and Scope lifecycle evidence. Kanban consumes these records for active Board and Focus views but must not redefine them. Every status change must retain stable Scope links and approval evidence so Kanban can reconcile flow without inventing state."
            },
            {
              "line": 274,
              "text": "## Pending ownership and governance corrections"
            },
            {
              "line": 276,
              "text": "Treat governance wording in an open documentation PR as **pending review evidence**, not as current default-branch policy. This applies to ownership, approval routing, authority boundaries, and lifecycle responsibilities."
            },
            {
              "line": 280,
              "text": "1. verify the current default-branch wording and the open PR's exact head separately;"
            },
            {
              "line": 282,
              "text": "3. allow downstream coordination artifacts to reference it as a dependency or blocker, but never as an already-effective decision;"
            },
            {
              "line": 283,
              "text": "4. preserve the established authority boundary while review is open\u2014for example, Scope owns product decisions and durable approval evidence, while General communicates project-wide implications;"
            },
            {
              "line": 288,
              "text": "## Synchronization Workflow"
            },
            {
              "line": 290,
              "text": "1. Confirm the exact project code and eight approved channel IDs."
            },
            {
              "line": 291,
              "text": "2. Read new messages from the other seven channels and enough `general` history to avoid duplicates."
            },
            {
              "line": 294,
              "text": "5. Inspect Scope navigation, Roadmap, Epics, Features, Ideas, task links, and lifecycle status."
            },
            {
              "line": 295,
              "text": "6. Reconcile repository state before editing: fetch the current default branch, read back open/merged PRs and exact-head CI, and remove or replace stale Roadmap review records. If reusing an existing coordination PR branch that another channel may also update, explicitly fetch that branch ref, compare local and remote ancestry, and fast-forward the clean worktree to the exact remote PR head before editing. Do not rely on a generic `fetch --prune` to create a missing remote-tracking ref, and never overwrite a concurrent Scope update. Never describe an already merged PR as open or carry an obsolete base SHA into a new coordination artifact. Follow [references/shared-coordination-pr-refresh.md](references/shared-coordination-pr-refresh.md) for the exact fetch, race-reconciliation, self-referential-head, remote-readback, and exact-head-CI procedure."
            },
            {
              "line": 296,
              "text": "7. Reconcile the Roadmap whenever current or predicted Epic/Feature state changed **or** cross-channel workflow evidence changed materially (for example, Architecture became merged, Kanban introduced `InIdeation`, Focus changed, or release-readiness evidence advanced). Replace stale PR head SHAs, exact-head CI task IDs, and channel-message evidence together; identify review ordering when one open coordination PR is a prerequisite for others. Keep Scope lifecycle status distinct from Kanban flow state: `InIdeation` does not imply `Approved for Solution`, and an empty Focus does not itself reject or defer scope. When the coordination artifact is inside the PR being refreshed, do not try to embed the commit's own final SHA: label the previous head/task as prior evidence, state that the refresh supersedes it, and put the newly pushed head plus matching terminal CI task in the synchronization report after remote verification."
            },
            {
              "line": 297,
              "text": "   - **Post-merge race reconciliation:** repository readback overrides channel summaries when PRs merge after the latest channel reports. If the shared coordination PR is already merged, do not push another update to its closed branch. Start a fresh review branch from the exact current default head; record each reviewed head, merge commit, current default SHA, and exact-SHA CI; remove obsolete open/pending wording from the Roadmap; and preserve all lifecycle and implementation gates. Inspect Board, Focus, Delivery, and release-readiness for the same stale review state. Update only General/Scope-owned coordination evidence directly, and route owner-specific artifact refreshes to Kanban, Delivery, or Releases unless the adopted governance explicitly authorizes shared editing. A documentation merge is evidence that wording became effective, not product, implementation, release, or deployment approval."
            },
            {
              "line": 300,
              "text": "10. Validate Scope menu order, dedicated sidebar hierarchy, identifiers, relationships, links, and the documentation build; verify changes through remote readback. For pushed review branches, verify PR head/base SHAs and require CI evidence matched to the exact pushed SHA rather than inheriting success from the base or another PR. If `corepack pnpm validate` fails only because nested package scripts cannot resolve `pnpm`, follow the `documentation-docusaurus-aerosim` package-manager fallback: run every command represented by the aggregate validator explicitly through `corepack pnpm` (including all repository-specific structural validators), then run the strict-link production build and `git diff --check`. Report the aggregate command failure as command-resolution context, not as a repository validation failure, and do not omit a nested validator while expanding the aggregate command."
            },
            {
              "line": 301,
              "text": "11. Immediately before reporting, perform a final incremental scan of only the eight authorized channels, starting after the per-channel high-watermarks observed earlier in the run. This closes the race in which a human decision, approval, release request, or another channel's completed coordination work arrives while validation or CI is running. Advance each high-watermark to the newest observed message ID, but cite the newest substantive message rather than a continuation-only verifier warning, delivery footer, or job-management boilerplate. If substantive evidence changed, reconcile the affected artifact and repository/PR facts again before reporting; do not publish a stale summary merely because the earlier push and CI succeeded."
            },
            {
              "line": 304,
              "text": "    - use exactly `[SILENT]` when the job contract requests that sentinel and there is genuinely no substantive result;"
            },
            {
              "line": 307,
              "text": "## Authority and Escalation"
            },
            {
              "line": 314,
              "text": "- run read-only checks;"
            },
            {
              "line": 318,
              "text": "Must request approval before:"
            },
            {
              "line": 320,
              "text": "- approving Scope, UI/UX, Architecture, Kanban, or release decisions in place of the assigned owner;"
            },
            {
              "line": 325,
              "text": "## Evidence Requirements"
            },
            {
              "line": 327,
              "text": "For completed work include exact evidence such as Discord message IDs, repository URLs, branch names, commit SHAs, documentation paths, CI run IDs, or check output. Mark uncertain conclusions as proposals."
            },
            {
              "line": 329,
              "text": "## Common Pitfalls"
            },
            {
              "line": 334,
              "text": "4. Proposing Epics or Features without user value, acceptance outcomes, evidence, or an owner."
            },
            {
              "line": 338,
              "text": "8. Placing Scope after Architecture or embedding it in another area's sidebar."
            },
            {
              "line": 339,
              "text": "9. Using generic `EPIC-0001`/`FEAT-0001` IDs instead of project-scoped `<CODE>-EP-N`/`<CODE>-FT-N` IDs."
            },
            {
              "line": 343,
              "text": "13. Conflating Scope lifecycle (`Proposed`, `Approved for Solution`, `Deferred`, `Rejected`) with Kanban flow (`InIdeation`, Focus states). Record both when useful, but never let a flow-state update silently alter lifecycle approval."
            },
            {
              "line": 344,
              "text": "14. Claiming CI for a coordination branch from the default branch, a stacked dependency, or another PR instead of matching a successful task to the exact pushed head SHA."
            },
            {
              "line": 347,
              "text": "## Verification Checklist"
            },
            {
              "line": 349,
              "text": "- [ ] Only the eight same-project channels were inspected."
            },
            {
              "line": 353,
              "text": "- [ ] Scope appears immediately before Architecture in the top menu and uses its own sidebar."
            },
            {
              "line": 354,
              "text": "- [ ] Scope sidebar order is Roadmap, Epics with nested Features, then Ideas."
            },
            {
              "line": 360,
              "text": "- [ ] No Epic or Feature was marked `Approved for Solution` without an explicit decision from the primary Scope owner or RootAtSkic as backup."
            },
            {
              "line": 362,
              "text": "- [ ] Completed activities include exact evidence."
            }
          ]
        }
      ],
      "remote_branch": "test",
      "remote_commit": "f97a4e8bbc1e107ddff0fac02a6730a32afed56e",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/productivity/corp-v1-channel-general-aerosim/SKILL.md"
    },
    {
      "skill": "corp-v1-channel-releases-aerosim",
      "repository": "corp-v1-aerosim-skills-code-agent/corp-v1-channel-releases-aerosim",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "feac09e2515486abc35a5b127525ec3a13ca97d47f12b32cf68e434c50df9ece",
          "lines": 355,
          "remote_sha256": "7df3df60acd4d8074c402ac93cc90e18b75a1e85caa3e59341b0e6d72247c8be",
          "matches_remote": false,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/corp-v1-channel-releases-aerosim/src/commit/3c4b4788915bae3d89303959c46a8854c9a141bc/SKILL.md",
          "excerpts": [
            {
              "line": 13,
              "text": "# AeroSim Releases Channel"
            },
            {
              "line": 15,
              "text": "<!-- AEROSIM_BLOCKER_OWNERSHIP:BEGIN -->"
            },
            {
              "line": 16,
              "text": "## Hermes owns blocker resolution"
            },
            {
              "line": 18,
              "text": "RootAtSkic's [project-wide correction](https://discord.com/channels/1518726359512387766/1537227683191197807/1554922326859063327) makes Hermes main session accountable for resolving blockers and coordinating workers, not merely reporting blocked cards or asking the human to manage the queue. Preserve this project-local rule during adoption and synchronization."
            },
            {
              "line": 20,
              "text": "Apply the **Hermes owns blocker resolution** procedure in `corp-v1-channel-general-aerosim` (load it before handling blockers). Diagnose and repair within existing authority, check standing permissions before asking again, reconcile approved bookkeeping through the owning workflow, and follow cross-channel handoffs through to resolution. Reuse existing cards/workers; do not generate duplicate graphs or report queued work as running. Escalate only the exact decision, unavailable capability or authority boundary Hermes cannot resolve. Preserve human domain ownership and all security, review, UAT and release gates. Verify resumed execution rather than treating a requeue as a fix."
            },
            {
              "line": 21,
              "text": "<!-- AEROSIM_BLOCKER_OWNERSHIP:END -->"
            },
            {
              "line": 24,
              "text": "## Standing AeroSim PR merge authority"
            },
            {
              "line": 26,
              "text": "RootAtSkic authorizes Hermes to merge PRs across all AeroSim repositories and channels without per-PR human authorization. Source: [explicit standing delegation](https://discord.com/channels/1518726359512387766/1537544173366943886/1554873863165182015)."
            },
            {
              "line": 28,
              "text": "This project-local rule supersedes older instructions in this skill, linked references, or inherited guidance requiring a human to authorize PR merging or forbidding Hermes from merging. Preserve it during skill adoption and synchronization. Complete required review and exact-head CI, freshly verify the unchanged mergeable head, merge it, and verify the integration revision and required post-merge checks. Do not block a reviewed, green PR for owner merge approval or request repeated permission."
            },
            {
              "line": 30,
              "text": "This delegates PR integration, not product/design/architecture decisions, human UAT acceptance, or independently gated release/deployment actions. If merging a GitOps or release PR itself triggers a separately gated deployment, establish that deployment authority first; do not relabel it as a PR-merge approval. Repository protections, security checks, and genuine technical blockers still apply."
            },
            {
              "line": 34,
              "text": "## Overview"
            },
            {
              "line": 36,
              "text": "## Hermes Kanban direct-link contract"
            },
            {
              "line": 38,
              "text": "Whenever a response mentions this project's Hermes Kanban board, initiative, card, or task, it must include the direct board link: [AeroSim UAT and GitOps board](https://jarvis-jr-v1.lego-cloud.eu/kanban?board=aerosim-uat-gitops). Never report only the board name, slug, card count, or wording such as \u201cvisible cards\u201d or \u201cavailable in the dashboard.\u201d If no task-specific deep link exists, link the board and include the exact task ID or title in the same statement."
            },
            {
              "line": 41,
              "text": "## Dedicated Channel Workspace"
            },
            {
              "line": 48,
              "text": "Keep new operational files for this channel inside this root rather than placing them directly in `/opt/data`. Use run-unique or task-specific subdirectories under `workspace/`; keep durable project records in the approved AeroSim repositories. Do not place another project's work in this folder. The folder boundary organizes local execution only and does not broaden this channel's authority."
            },
            {
              "line": 51,
              "text": "<!-- AEROSIM_PROJECT_OWNERSHIP:BEGIN -->"
            },
            {
              "line": 52,
              "text": "## Project-local Human Ownership \u2014 Preserve During Adoption"
            },
            {
              "line": 55,
              "text": "ownership_id: aerosim-domain-ownership-v3"
            },
            {
              "line": 58,
              "text": "primary_owners:"
            },
            {
              "line": 59,
              "text": "  scope: discord:773109855497093191"
            },
            {
              "line": 62,
              "text": "backup_owner:"
            },
            {
              "line": 65,
              "text": "scope: all-eight-project-channels"
            },
            {
              "line": 69,
              "text": "The named human is the primary accountable owner and routine decision/approval authority for that domain. Scope decisions route to `<@773109855497093191>`, UI/UX decisions route to Dovydas (`<@1543952051212656684>`), and Architecture decisions route to `<@1543940246071410706>`. RootAtSkic (`<@1518725627845283888>`) retains backup ownership across every AeroSim channel: RootAtSkic may act when a primary owner is unavailable, when work is explicitly escalated, to resolve cross-domain conflict, or to override a domain decision."
            },
            {
              "line": 71,
              "text": "No new primary owner is assigned here for General, Kanban, Delivery, UAT, or Releases; RootAtSkic retains accountable ownership for those domains. Hermes remains the operational and coordination agent: it may research, draft, implement approved work, validate, publish, merge, release, deploy, and verify only where the owning channel workflow already authorizes those actions. Hermes must not substitute itself for the named Scope, UI/UX, or Architecture owner when an exact domain decision or approval is required unless RootAtSkic explicitly delegates that exact action."
            },
            {
              "line": 73,
              "text": "RootAtSkic's 2026-09-28 decision supersedes ownership version 2 for Scope only: Jone Tamulaite - Nong left AeroSim and MartynasP (`<@773109855497093191>`) became the primary Scope owner. Preserve earlier decisions as history; route new Scope decisions to MartynasP."
            },
            {
              "line": 75,
              "text": "This decision supersedes the earlier broad Hermes project-lead delegation only where it conflicts with the three named human domain owners. It does not waive exact-item decision records, canonical evidence, accessibility review, CI, security/quality gates, external permissions, spending controls, legal commitments, or confirmation for destructive or irreversible high-impact operations. This project-local block must be preserved during central/reference skill adoption."
            },
            {
              "line": 76,
              "text": "<!-- AEROSIM_PROJECT_OWNERSHIP:END -->"
            },
            {
              "line": 80,
              "text": "This skill owns release readiness and the controlled release-change workflow for AeroSim. It monitors the project's `general`, `scope`, `architecture`, `ui-ux`, `kanban`, `delivery`, `uat`, and `releases` channels for release implications. Release execution requires both an explicit request from RootAtSkic and verified `UAT_PASSED` evidence bound to the exact development deployment tuple being promoted. A UAT pass does not itself authorize release, and a release request does not waive UAT."
            },
            {
              "line": 82,
              "text": "Load the project-adopted `documentation-docusaurus-aerosim` skill from `https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/documentation-docusaurus-aerosim` before changing documentation structure, navigation, Markdown/MDX, Docusaurus configuration, data-backed React components, or builds."
            },
            {
              "line": 96,
              "text": "The application and E2E repositories are not deployable manifest sources. Delivery may update only development desired state; Releases owns production promotion."
            },
            {
              "line": 98,
              "text": "The skill opens a pull request for review. Deployment must not proceed until RootAtSkic, as retained Releases owner, approves the exact pull request and all required gates pass."
            },
            {
              "line": 100,
              "text": "Load the global `corp-v1-glossary` skill from `https://gitea.lego-cloud.eu/home-v1-skills-code-agent/corp-v1-glossary` whenever project documentation needs to define or explain a reusable term. Maintain one canonical definition in the project's final top-level **Glossary** area and link to it from the owning domain page; do not duplicate glossary-style explanations across channel documentation."
            },
            {
              "line": 102,
              "text": "## Shared Corp v1 System Login"
            },
            {
              "line": 104,
              "text": "When an authorized task requires login to a Corp v1 system being built or operated, follow the shared policy in `corp-v1-main` and use only the Bitwarden-injected runtime secrets named:"
            },
            {
              "line": 111,
              "text": "Secret availability is capability, not authorization. Verify the destination origin and task purpose before login. Never print, inspect, log, hash, serialize, paste, screenshot, or persist either value; never place a value in a command line, URL, file, repository, prompt, Discord message, browser console, test fixture, CI output, or generated artifact. Never ask a human to paste a value into chat. If a variable is unavailable, report only its missing name and request Bitwarden/gateway injection. Login does not authorize account recovery, MFA or credential changes, permission changes, billing, spending, destructive operations, or access outside the approved project task."
            },
            {
              "line": 113,
              "text": "## When to Use"
            },
            {
              "line": 118,
              "text": "- running its recurring synchronization job;"
            },
            {
              "line": 127,
              "text": "## Proactive Iteration Requirement"
            },
            {
              "line": 131,
              "text": "## Approved Information Boundary"
            },
            {
              "line": 133,
              "text": "Inspect only:"
            },
            {
              "line": 137,
              "text": "corp-v1-aerosim-scope"
            },
            {
              "line": 146,
              "text": "Use mapped-channel history to avoid duplicate work and new relevant activity from the other seven channels. Never inspect another project's channels."
            },
            {
              "line": 148,
              "text": "## Cumulative Scope Version Contract"
            },
            {
              "line": 150,
              "text": "Before release readiness analysis, resolve the canonical Release parent and its complete versioned Epic/Feature hierarchy from `data/releases/`:"
            },
            {
              "line": 156,
              "text": "The first Release includes initial Epics and Features at `v1`. Every later Release carries every prior entry. `UNCHANGED` keeps the prior version; `UPDATED` increments it by one and requires both a change summary and a difference-from-previous explanation; `NEW` starts at `v1`. A new or updated Feature requires its owning Epic to be `NEW` or `UPDATED`."
            },
            {
              "line": 158,
              "text": "Release readiness must compare the complete cumulative hierarchy, not only newly delivered Tasks. Preserve the original release allocation and immutable evidence for prior Tasks. Add new Tasks for a changed delta or new capability rather than duplicating or rewriting completed prior-release Tasks. Fail closed when a later Release omits prior scope, skips a version, repeats an ID, maps a Feature to the wrong Epic, or marks an Epic unchanged while its child Feature changed."
            },
            {
              "line": 160,
              "text": "## Release Readiness"
            },
            {
              "line": 162,
              "text": "Before proposing or executing a release, verify:"
            },
            {
              "line": 164,
              "text": "- explicit release request from RootAtSkic and intended scope;"
            },
            {
              "line": 167,
              "text": "- Kanban `ToBeReleased` evidence for included Epic/Feature/task scope;"
            },
            {
              "line": 172,
              "text": "- rollback strategy and previous known-good state;"
            },
            {
              "line": 173,
              "text": "- deployment verification and ownership plan;"
            },
            {
              "line": 178,
              "text": "## Controlled Gondor v1 Release Workflow"
            },
            {
              "line": 180,
              "text": "### Gate 1 \u2014 explicit request"
            },
            {
              "line": 182,
              "text": "Require an unambiguous request from RootAtSkic identifying the project and intended release. If version, environment, artifact, or scope is missing, prepare a readiness summary and ask for the missing decision rather than guessing."
            },
            {
              "line": 184,
              "text": "### Gate 2 \u2014 discover repositories and instructions"
            },
            {
              "line": 189,
              "text": "4. Identify the exact project-owned template subtree and generated destination."
            },
            {
              "line": 192,
              "text": "### Gate 3 \u2014 update template source"
            },
            {
              "line": 194,
              "text": "Make the smallest project-scoped change in `home-v1/gondor-v1-tmpl`. The template repository remains the source of truth; do not patch generated Gondor output manually as a substitute."
            },
            {
              "line": 198,
              "text": "### Gate 4 \u2014 render"
            },
            {
              "line": 202,
              "text": "- only expected project files changed;"
            },
            {
              "line": 209,
              "text": "### Gate 5 \u2014 separate branch in Gondor v1"
            },
            {
              "line": 217,
              "text": "Use the project's actual branching convention when defined. Commit only intended rendered artifacts, push the branch, and verify remote SHA equality and readback."
            },
            {
              "line": 221,
              "text": "### Gate 6 \u2014 pull request"
            },
            {
              "line": 225,
              "text": "- project, environment, version, and requested scope;"
            },
            {
              "line": 233,
              "text": "- explicit statement: **RootAtSkic approval required before deployment**."
            },
            {
              "line": 237,
              "text": "### Gate 7 \u2014 stop before deployment"
            },
            {
              "line": 239,
              "text": "Complete reviewed PR integration under standing AeroSim merge authority. GitOps synchronization or deployment still requires the applicable release/deployment authorization; when merging itself triggers deployment, establish that authority before merging."
            },
            {
              "line": 241,
              "text": "## Post-Approval and Deployment Evidence"
            },
            {
              "line": 245,
              "text": "- verify approval and required checks still apply to the exact head SHA;"
            },
            {
              "line": 247,
              "text": "- inspect `0600-aerosim` for Argo tracking/ownership metadata before rollout. It is tracked by `0000-aerosim-app-of-apps` (tracking ID `0000-aerosim-app-of-apps:argoproj.io/Application:argocd/0600-aerosim`). After the reviewed GitOps merge, let the owning app-of-apps reconcile; a hard refresh of that root Application can request immediate discovery. Do not directly apply or patch the child Application. Verify the root's exact merged Git revision, the child's chart/image-source revision and Synced/Healthy state, and pod `imageID` digests;"
            },
            {
              "line": 250,
              "text": "- execute browser acceptance through the same visible controls and ordinary interactions available to a human; a forced click, direct DOM/state mutation, hidden control, or API bypass invalidates that journey as acceptance evidence and may be used only for separately labelled diagnosis;"
            },
            {
              "line": 257,
              "text": "Never claim deployment from a merged PR alone. A technically complete deployment can remain unaccepted when human confirmation is absent, and a reported blocking defect changes human acceptance to failed until a verified forward fix is released and accepted."
            },
            {
              "line": 259,
              "text": "## Production Rejection and Replacement"
            },
            {
              "line": 261,
              "text": "A deployed release that makes a core user journey unusable, requires test-only interaction bypasses, or exposes a critical runtime failure is **rejected for production**. Healthy pods, green CI, clean browser-console output, or successful route loading do not make that release suitable for continued production designation."
            },
            {
              "line": 266,
              "text": "2. choose containment with the Releases owner: roll back to the previous verified release when one exists, or place the affected experience in maintenance/unavailable state when no safe prior release exists; do not keep an unusable core journey publicly presented as production merely because infrastructure is healthy;"
            },
            {
              "line": 269,
              "text": "5. require ordinary-user-path automated acceptance plus explicit human post-release confirmation before designating the replacement as accepted production;"
            },
            {
              "line": 274,
              "text": "## Release Completion Gate \u2014 Mark Tasks `DONE`"
            },
            {
              "line": 276,
              "text": "This gate belongs only to the AeroSim Releases process. Delivery hands off Tasks whose files say `TO_BE_RELEASED`; Releases must not perform this transition for a Task still in another state."
            },
            {
              "line": 280,
              "text": "**Runtime:** Hermes main session, Gitea Actions, and Gondor/GitOps operate only within the explicitly approved release scope. Cron or HermesSubAgent must not silently start or complete a release."
            },
            {
              "line": 284,
              "text": "**Exit gate:** The approved deployment and technical release closure are complete, its runtime result and release-specific automated acceptance checks are verified, the Task documentation change is merged, and each included Task file on remote documentation `test` says `DONE`. Record human post-release acceptance separately as `PENDING`, `PASSED`, or `FAILED`; do not convert technical closure into human acceptance. A human-reported blocking defect sets human acceptance to `FAILED` and requires a verified forward fix plus new explicit human confirmation."
            },
            {
              "line": 286,
              "text": "## Synchronization Workflow"
            },
            {
              "line": 288,
              "text": "1. Read new same-project activity from **all eight allowed channels** (`general`, `scope`, `architecture`, `ui-ux`, `kanban`, `delivery`, and `releases`) and enough release history to avoid duplicates. Treat a reader or helper that omits Scope, UI/UX, or Kanban as incomplete; update it before assessing readiness."
            },
            {
              "line": 290,
              "text": "3. Reconcile Kanban explicitly on every substantive readiness refresh: record whether exact items exist in `ToBeReleased`, distinguish an open Kanban PR from merged default-branch evidence, and never infer release scope from green Kanban/documentation CI. An empty Focus or `ToBeReleased` view is a concrete release blocker worth recording when newly established or changed."
            },
            {
              "line": 291,
              "text": "4. Without an explicit release request, proactively improve readiness documentation, version/changelog evidence, rollback planning, or a concrete blocker/approval request only. When previously open readiness, Scope, Architecture, Delivery, or Kanban PRs have merged, replace stale open/stacked-PR language with the verified merge chain, final default-branch SHA, and exact-head CI evidence; preserve that documentation CI is not application release evidence. If another documentation PR remains open, state its exact head, base, merge state, and exact-head CI independently rather than folding it into the merged baseline."
            },
            {
              "line": 292,
              "text": "5. Put autonomous readiness-documentation improvements on a dedicated branch and open a PR, complete required review and exact-head CI, then merge it under standing AeroSim authority and verify post-merge checks. If an open release-readiness PR already owns the same artifact and purpose, refresh that branch and PR in place instead of creating another overlapping PR. If the prior readiness PR is closed or merged, treat it as historical evidence: branch a successor from the freshly fetched default branch rather than reviving or pushing the old source branch. Do not stack the readiness page onto open Scope, Architecture, Kanban, or Delivery branches merely because they contain related evidence; their artifacts and review lifecycles remain independent. Verify local validation, exact remote ref equality, authenticated artifact readback, PR head/base SHAs, and exact-head CI before reporting success. Follow [references/recurring-readiness-pr-refresh.md](references/recurring-readiness-pr-refresh.md) for the reusable discovery, replacement-map, race-guard, publication, and authenticated readback procedure. Keep reusable collectors and verifiers parameterized; do not hard-code a run's PR numbers, SHAs, or task IDs into durable helpers."
            },
            {
              "line": 293,
              "text": "6. Guard against cross-channel evidence races: after reading all eight channels, re-read the exact Gitea PR heads and exact-head CI for any Scope, Architecture, UI/UX, Kanban, Delivery, or readiness evidence cited in the document immediately before editing. A channel report or earlier API snapshot may already be stale because another synchronization job advanced its branch. Replace every occurrence of the superseded SHA/task pair in the durable artifact, not only the summary row. Also scan for stale boundary language such as an obsolete channel count (for example, \u201cfour channels\u201d after Scope and Kanban were added)."
            },
            {
              "line": 294,
              "text": "7. Treat CI task identity as mutable even when a PR head SHA is unchanged: retries or later workflow runs can produce multiple tasks for the same exact SHA. Re-read the task list immediately before editing, select the latest terminal task/run by numeric ID or creation time, require its successful status, and update every stale task reference paired with that SHA. Do not report an older successful task merely because the commit did not change."
            },
            {
              "line": 295,
              "text": "8. Add a **pre-commit race guard** after local validation and immediately before committing: re-read every cited Scope, Architecture, Kanban, Delivery, and readiness PR plus the latest exact-head CI task. Compare these fresh pairs with the edited artifact. If any cited head or latest task advanced while validation ran, update every occurrence, rerun the affected validation, and repeat the guard. Do not commit a page already known to contain stale evidence."
            },
            {
              "line": 296,
              "text": "9. After pushing a refreshed readiness branch, poll CI for the new exact SHA, read back the artifact through the authenticated API, and re-read the PR. In the authenticated readback, assert that every newly cited PR-head/task pair is present and every superseded SHA/task pair is absent; a distinctive heading alone is insufficient evidence that the whole page was reconciled. Re-read the task list once more immediately before the final report because a retry can change the latest task identity after the first successful poll. Report only the new head/latest-task pair; never carry forward the prior head's successful task as evidence for the updated PR."
            },
            {
              "line": 297,
              "text": "   - Do not embed the readiness PR's own mutable head SHA or exact-head task as durable \u201ccurrent evidence\u201d inside the artifact it owns. Every commit would immediately invalidate that self-reference and cause an endless refresh loop. Cite mutable heads/tasks for other evidence PRs when needed; verify and report the readiness PR's own final head/task externally after push."
            },
            {
              "line": 302,
              "text": "    - use exactly `[SILENT]` when the job contract requests that sentinel and there is genuinely no substantive result;"
            },
            {
              "line": 305,
              "text": "## Authority and Escalation"
            },
            {
              "line": 310,
              "text": "- update project-scoped template configuration;"
            },
            {
              "line": 316,
              "text": "Requires separate approval from RootAtSkic, as retained Releases owner, for:"
            },
            {
              "line": 325,
              "text": "## Common Pitfalls"
            },
            {
              "line": 337,
              "text": "11. Creating a second readiness PR when an open PR already owns the same page and purpose, increasing review conflicts and stale evidence."
            },
            {
              "line": 339,
              "text": "13. Treating an earlier channel report as current evidence without re-reading the referenced PR head and exact-head CI immediately before the documentation update."
            },
            {
              "line": 341,
              "text": "15. Running the Copier wrapper from a checkout containing an ignored `.venv/`: the wrapper copies that environment into its disposable source tree and Copier can fail with `ForbiddenPathError` on `.venv/bin/python3`. Use a Copier executable installed outside the template checkout, remove only the ignored local `.venv/`, render into an absent/empty staging directory, and then publish only the generator-owned paths after reviewing unrelated drift."
            },
            {
              "line": 343,
              "text": "## Verification Checklist"
            },
            {
              "line": 345,
              "text": "- [ ] RootAtSkic explicitly requested the release."
            },
            {
              "line": 346,
              "text": "- [ ] Scope, version, environment, artifact, and source SHA are known."
            },
            {
              "line": 348,
              "text": "- [ ] Only the project's template portion changed."
            },
            {
              "line": 350,
              "text": "- [ ] Rendered output contains only expected changes."
            },
            {
              "line": 353,
              "text": "- [ ] The PR clearly requires RootAtSkic approval before deployment."
            }
          ]
        }
      ],
      "remote_branch": "test",
      "remote_commit": "3c4b4788915bae3d89303959c46a8854c9a141bc",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/productivity/corp-v1-channel-releases-aerosim/SKILL.md"
    },
    {
      "skill": "home-v1-discord",
      "repository": "home-v1-skills-code-agent/home-v1-discord",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "e3e8f5303a6dcca1e6d0f8b2c2da6b179f6eb8de81608f20b7c825c0a3506061",
          "lines": 353,
          "remote_sha256": "49c8f4568b216fe1f4c3a60901c7f8b2a2eabc673d895764c229efb3f5ba52fa",
          "matches_remote": false,
          "remote_url": "https://gitea.lego-cloud.eu/home-v1-skills-code-agent/home-v1-discord/src/commit/910c4b615aa64e9645d50df14d9903c08a0e1a1f/SKILL.md",
          "excerpts": [
            {
              "line": 14,
              "text": "# Home v1 Discord Operations"
            },
            {
              "line": 16,
              "text": "## Overview"
            },
            {
              "line": 18,
              "text": "This skill is the reusable operational playbook for Discord administration in the Home v1 environment. It owns the mechanics for Discord discovery, provisioning, permissions, messages, pins, scheduled delivery, deletion, and verification."
            },
            {
              "line": 20,
              "text": "Domain skills such as `corp-v1-steering-committee` retain their own policy: which channels are allowed, their names and purposes, who approves them, what must be posted, and what lifecycle rules apply. They reference this skill for execution rather than duplicating Discord API or tool procedures."
            },
            {
              "line": 22,
              "text": "A tool limitation in one delegated or scheduled execution context is not evidence that Hermes globally lacks Discord administration capability. Report the scope accurately, return to a context with the required tool when possible, and use the configured REST fallback only when necessary."
            },
            {
              "line": 24,
              "text": "## When to Use"
            },
            {
              "line": 33,
              "text": "- create or validate channel-targeted recurring jobs;"
            },
            {
              "line": 37,
              "text": "Do not use it to decide a project\u2019s channel topology or governance policy. The calling domain skill owns those decisions."
            },
            {
              "line": 39,
              "text": "## Authoritative Environment"
            },
            {
              "line": 43,
              "text": "- Use Discord REST API v10 only when the active context lacks the required administration operation and an authenticated gateway credential is already configured."
            },
            {
              "line": 46,
              "text": "## Execution Context and Capability Rules"
            },
            {
              "line": 48,
              "text": "Before claiming that an operation is impossible:"
            },
            {
              "line": 54,
              "text": "5. If the native operation is unavailable, check only for the **name/presence** of an already configured Discord credential; do not print its value."
            },
            {
              "line": 55,
              "text": "6. Use the REST fallback from the trusted main environment when authorized and technically necessary. A terminal backend may lack Discord environment variables even though the active gateway profile has a configured credential. When `execute_code` is available, its direct Python context can differ from `hermes_tools.terminal`: check credential-name presence and the active `HERMES_HOME` there without dumping environment/configuration values. If the active profile's existing `.env` contains `DISCORD_BOT_TOKEN`, load only that value into memory with `dotenv_values(Path(os.environ.get('HERMES_HOME', '/opt/data')) / '.env').get('DISCORD_BOT_TOKEN')` on this deployment and perform bounded Discord REST calls directly in that Python context. Never print or persist the token, change credential injection, inspect another profile, or return headers. Verify `/users/@me`, the exact authorized channel/message, and a meaningful authorized send plus separate readback before declaring read/send access restored. Pass the non-secret method and verified IDs to the same blocked worker; do not merely requeue it without resolving the path."
            },
            {
              "line": 56,
              "text": "7. If neither path exists, report the exact missing capability and stop without fabricating IDs or results."
            },
            {
              "line": 66,
              "text": "## Safety and Authorization"
            },
            {
              "line": 68,
              "text": "1. Obtain authorization from the calling workflow before mutating Discord state."
            },
            {
              "line": 69,
              "text": "2. Discover before mutation and make operations idempotent."
            },
            {
              "line": 70,
              "text": "3. Limit changes to the exact guild, category, channels, identities, messages, and permissions approved."
            },
            {
              "line": 75,
              "text": "8. Treat channel deletion and broad permission changes as destructive actions requiring explicit scope."
            },
            {
              "line": 77,
              "text": "10. Preserve exact partial state if a multi-step operation fails, then resume idempotently."
            },
            {
              "line": 79,
              "text": "## Discovery Procedure"
            },
            {
              "line": 81,
              "text": "Record the approved target names and expected counts before changing anything."
            },
            {
              "line": 88,
              "text": "- exact name collisions anywhere in the guild;"
            },
            {
              "line": 93,
              "text": "- scheduled jobs already targeting the channels."
            },
            {
              "line": 95,
              "text": "For a name-based request, match both the exact name and expected resource type. Reject ambiguity rather than choosing the first result."
            },
            {
              "line": 97,
              "text": "When a domain workflow requires exactly N channels matching a prefix, inspect the full guild channel list, not only the expected category. An identically named or prefixed channel elsewhere is still a collision."
            },
            {
              "line": 99,
              "text": "## Identity Resolution"
            },
            {
              "line": 103,
              "text": "1. Search guild membership by exact known username, global display name, or supplied ID."
            },
            {
              "line": 108,
              "text": "6. Do not map Discord names to Gitea or other platform usernames without independent verification."
            },
            {
              "line": 110,
              "text": "## Permission Model"
            },
            {
              "line": 112,
              "text": "Evaluate effective channel access, not only guild-wide or category-level settings."
            },
            {
              "line": 122,
              "text": "- Manage Channels only when Hermes must create, move, update, lock, or delete channels."
            },
            {
              "line": 126,
              "text": "1. guild owner/administrator state;"
            },
            {
              "line": 133,
              "text": "Do not add a member-specific overwrite if an existing role already grants the required effective access unless explicit isolation is needed. Do not rewrite the complete overwrite list merely to add one permission."
            },
            {
              "line": 135,
              "text": "Verify operationally where harmless:"
            },
            {
              "line": 142,
              "text": "- list pins and confirm the exact message ID."
            },
            {
              "line": 146,
              "text": "## Idempotent Channel Provisioning"
            },
            {
              "line": 148,
              "text": "Given an approved category and exact channel definitions:"
            },
            {
              "line": 153,
              "text": "   - reuse an exact matching text channel in the correct category;"
            },
            {
              "line": 154,
              "text": "   - reject an exact-name resource of the wrong type;"
            },
            {
              "line": 156,
              "text": "   - create only when no matching resource exists."
            },
            {
              "line": 157,
              "text": "4. Apply only approved topic, position, slow-mode, NSFW, archive, or overwrite settings."
            },
            {
              "line": 160,
              "text": "7. Assert exact names, types, category placement, and required count or prefix constraints."
            },
            {
              "line": 163,
              "text": "A rerun must not duplicate channels or guidance messages. Use a distinctive marker, known message ID, or exact content check to detect already provisioned guidance."
            },
            {
              "line": 165,
              "text": "## Messages and Pins"
            },
            {
              "line": 170,
              "text": "2. Post only after the channel ID and effective send permission are verified."
            },
            {
              "line": 172,
              "text": "4. Pin only when required."
            },
            {
              "line": 173,
              "text": "5. List pinned messages and confirm the exact ID."
            },
            {
              "line": 175,
              "text": "7. On updates, edit or replace the known guidance message rather than posting duplicates."
            },
            {
              "line": 179,
              "text": "## Native Tool Path"
            },
            {
              "line": 190,
              "text": "After each mutation, perform a separate read. Batch independent reads, but serialize operations when later actions depend on newly created IDs."
            },
            {
              "line": 194,
              "text": "## Authenticated REST API Fallback"
            },
            {
              "line": 202,
              "text": "The runtime credential is commonly exposed as `DISCORD_BOT_TOKEN`, but inspect configured environment variable names rather than assuming. Check presence without printing values."
            },
            {
              "line": 204,
              "text": "Construct authorization only inside the process:"
            },
            {
              "line": 242,
              "text": "For channel creation, submit only approved fields, for example:"
            },
            {
              "line": 255,
              "text": "## Temporary Scripts"
            },
            {
              "line": 265,
              "text": "- use exact guild/category/channel constants from approved scope;"
            },
            {
              "line": 266,
              "text": "- implement GET-before-POST and readback verification;"
            },
            {
              "line": 267,
              "text": "- output only non-secret resource IDs and verification facts;"
            },
            {
              "line": 272,
              "text": "## Channel-Targeted Scheduled Jobs"
            },
            {
              "line": 274,
              "text": "The calling domain skill defines whether jobs are required, their review lens, cadence, and channel boundaries. This skill owns safe creation and verification mechanics."
            },
            {
              "line": 278,
              "text": "1. Wait until target channel IDs are known and verified."
            },
            {
              "line": 280,
              "text": "3. Include exact guild ID, allowed channel IDs and names, target channel, purpose, evidence expectations, and silence behavior."
            },
            {
              "line": 283,
              "text": "6. Deliver to the exact target channel or thread, preserving topic context."
            },
            {
              "line": 285,
              "text": "8. List jobs after creation and verify ID, enabled status, cadence, toolsets, prompt scope, and delivery target."
            },
            {
              "line": 287,
              "text": "10. At teardown, pause and remove jobs before deleting their target channels, then verify they no longer exist."
            },
            {
              "line": 291,
              "text": "## Destructive Changes and Teardown"
            },
            {
              "line": 293,
              "text": "For deletion or lockdown:"
            },
            {
              "line": 295,
              "text": "1. Confirm explicit scope and authorization."
            },
            {
              "line": 298,
              "text": "4. Post any approved closure notice before access is removed."
            },
            {
              "line": 300,
              "text": "6. Delete only exact approved channel IDs, never all prefix matches without review."
            },
            {
              "line": 305,
              "text": "## Verification Record"
            },
            {
              "line": 322,
              "text": "Never claim success based only on local script completion, an HTTP 2xx mutation, or a job creation response. Read the resulting remote state independently."
            },
            {
              "line": 324,
              "text": "## Common Pitfalls"
            },
            {
              "line": 329,
              "text": "4. Creating before checking the complete guild inventory."
            },
            {
              "line": 334,
              "text": "9. Persisting credentials in scripts, repositories, cron prompts, or messages."
            },
            {
              "line": 336,
              "text": "11. Creating scheduled jobs before channel IDs are known."
            },
            {
              "line": 338,
              "text": "13. Deleting channels before preserving required history and removing jobs."
            },
            {
              "line": 341,
              "text": "## Verification Checklist"
            },
            {
              "line": 343,
              "text": "- [ ] Authorization and exact scope are known."
            },
            {
              "line": 345,
              "text": "- [ ] Guild, category, channels, identities, roles, and existing jobs were discovered before mutation."
            },
            {
              "line": 346,
              "text": "- [ ] Exact names, types, category placement, and collision rules were enforced."
            },
            {
              "line": 347,
              "text": "- [ ] Only approved permission changes were applied."
            },
            {
              "line": 353,
              "text": "- [ ] The calling workflow received stable IDs and exact remaining blockers."
            }
          ]
        }
      ],
      "remote_branch": "test",
      "remote_commit": "910c4b615aa64e9645d50df14d9903c08a0e1a1f",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/software-development/home-v1-discord/SKILL.md"
    },
    {
      "skill": "documentation-docusaurus-aerosim",
      "repository": "corp-v1-aerosim-skills-code-agent/documentation-docusaurus-aerosim",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "9a650f040842f17b3e72a2c655b68f40e8aaa847157fdd8ae4e33191582d1409",
          "lines": 222,
          "remote_sha256": "f1bd86adebc24d092d6cc088cac7343518e52f0d2175b64609d8d69d61578204",
          "matches_remote": false,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/documentation-docusaurus-aerosim/src/commit/2e1961a0850d17d94d855c129183b9d026dd90f7/SKILL.md",
          "excerpts": [
            {
              "line": 8,
              "text": "  Requires the intended reader and outcome before writing, keeps documentation types separate, and"
            },
            {
              "line": 22,
              "text": "# Docusaurus Documentation Skill"
            },
            {
              "line": 25,
              "text": "## Standing AeroSim PR merge authority"
            },
            {
              "line": 27,
              "text": "RootAtSkic authorizes Hermes to merge PRs across all AeroSim repositories and channels without per-PR human authorization. Source: [explicit standing delegation](https://discord.com/channels/1518726359512387766/1537544173366943886/1554873863165182015)."
            },
            {
              "line": 29,
              "text": "This project-local rule supersedes older instructions in this skill, linked references, or inherited guidance requiring a human to authorize PR merging or forbidding Hermes from merging. Preserve it during skill adoption and synchronization. Complete required review and exact-head CI, freshly verify the unchanged mergeable head, merge it, and verify the integration revision and required post-merge checks. Do not block a reviewed, green PR for owner merge approval or request repeated permission."
            },
            {
              "line": 31,
              "text": "This delegates PR integration, not product/design/architecture decisions, human UAT acceptance, or independently gated release/deployment actions. If merging a GitOps or release PR itself triggers a separately gated deployment, establish that deployment authority first; do not relabel it as a PR-merge approval. Repository protections, security checks, and genuine technical blockers still apply."
            },
            {
              "line": 37,
              "text": "## My Philosophy"
            },
            {
              "line": 41,
              "text": "I follow the **Di\u00e1taxis framework**. Before writing anything, identify which quadrant you're in:"
            },
            {
              "line": 52,
              "text": "## Before You Write: Questions I'll Ask"
            },
            {
              "line": 59,
              "text": "4. **What do they already know?** (Prerequisites matter)"
            },
            {
              "line": 63,
              "text": "## Structure: My Strong Opinions"
            },
            {
              "line": 65,
              "text": "### Sidebar Organization"
            },
            {
              "line": 77,
              "text": "\u2514\u2500\u2500 resources/                # Links, community, changelog"
            },
            {
              "line": 82,
              "text": "### Frontmatter: Non-Negotiables"
            },
            {
              "line": 94,
              "text": "### Writing Rules I Enforce"
            },
            {
              "line": 98,
              "text": "- One path only\u2014no \"alternatively\" or \"you could also\""
            },
            {
              "line": 109,
              "text": "- Mirror the code structure exactly"
            },
            {
              "line": 120,
              "text": "## Inclusive Language: Required"
            },
            {
              "line": 137,
              "text": "## Dynamic Documentation Patterns"
            },
            {
              "line": 139,
              "text": "### Local YAML data-backed documentation"
            },
            {
              "line": 141,
              "text": "When Epics, Features, Tasks, requirements, or architectural decisions must feed several documentation views, keep the authoritative records in a local `.yml`-only data tree and load them during the Docusaurus build. Requirements must use separate functional and non-functional folders, and architectural decisions must use the exact `data/architecture-decisions/` folder. Do not duplicate records inside MDX. Follow [references/local-yaml-data-database.md](references/local-yaml-data-database.md) for record schemas, stable IDs, canonical relationships, generated indexes, validation, React hooks/components, MDX composition, and maintenance commands."
            },
            {
              "line": 143,
              "text": "For a cumulative release model, keep Release parents in `data/releases/` and reference canonical Epic and Feature IDs rather than copying their records. Validate prior-release carry-forward, exact version increments, parentage, and unique membership. Render separate reader fields for what changed and how the new version differs from the prior version. Unchanged entries retain their version and must not display invented change prose."
            },
            {
              "line": 145,
              "text": "### React Flow roadmaps and dependency maps"
            },
            {
              "line": 147,
              "text": "Use `@xyflow/react` when validated documentation data needs an interactive roadmap, Gantt-like timeline, Feature dependency graph, Requirement-to-Feature traceability map, or architectural decision-to-Feature map. React Flow is the renderer, not the source database or an automatic layout engine. Follow [references/react-flow-visualizations.md](references/react-flow-visualizations.md) for Docusaurus BrowserOnly integration, deterministic Gantt positioning, Dagre/ELK layout choices, source-YAML adapters, read-only interaction, accessibility, sharp styling, semantic table fallbacks, testing, and browser verification."
            },
            {
              "line": 149,
              "text": "### Draw.io source diagrams"
            },
            {
              "line": 153,
              "text": "### Interactive Elements (Use Sparingly)"
            },
            {
              "line": 171,
              "text": "### Admonitions: The Hierarchy"
            },
            {
              "line": 182,
              "text": "## What I Won't Do"
            },
            {
              "line": 191,
              "text": "## Reference Files"
            },
            {
              "line": 206,
              "text": "## Package-manager validation pitfall"
            },
            {
              "line": 208,
              "text": "If `pnpm` is available only through `corepack`, `corepack pnpm validate` can still fail when the `validate` script invokes nested `pnpm` commands, because child scripts resolve `pnpm` from `PATH`. Before treating this as a repository failure, either:"
            },
            {
              "line": 215,
              "text": "## Getting Started"
            }
          ]
        },
        {
          "path": "references/writing-guide.md",
          "installed_sha256": "f6ccbcc6e0856735bd7e229f00fc8067611cddefde67dae1bf58ae0b9a39e1b8",
          "lines": 276,
          "remote_comparison": "not performed for this supplemental file",
          "excerpts": [
            {
              "line": 1,
              "text": "# Writing Guide"
            },
            {
              "line": 7,
              "text": "## Voice Principles"
            },
            {
              "line": 9,
              "text": "### Be Direct"
            },
            {
              "line": 17,
              "text": "### Address the Reader"
            },
            {
              "line": 25,
              "text": "### Present Tense, Active Voice"
            },
            {
              "line": 35,
              "text": "## Words That Weaken"
            },
            {
              "line": 49,
              "text": "**Before:** \"Simply run the following command to easily set up...\""
            },
            {
              "line": 54,
              "text": "## Inclusive Language"
            },
            {
              "line": 56,
              "text": "### Required Replacements"
            },
            {
              "line": 74,
              "text": "### Writing About People"
            },
            {
              "line": 88,
              "text": "### Cultural Sensitivity"
            },
            {
              "line": 97,
              "text": "## Second Person \"You\""
            },
            {
              "line": 108,
              "text": "- Only when the reader is genuinely working WITH you"
            },
            {
              "line": 114,
              "text": "## Code in Prose"
            },
            {
              "line": 116,
              "text": "### Commands"
            },
            {
              "line": 123,
              "text": "### Filenames and Paths"
            },
            {
              "line": 127,
              "text": "### Configuration Values"
            },
            {
              "line": 131,
              "text": "### Placeholders"
            },
            {
              "line": 143,
              "text": "## Headings"
            },
            {
              "line": 145,
              "text": "### Capitalization"
            },
            {
              "line": 152,
              "text": "### Parallel Structure"
            },
            {
              "line": 166,
              "text": "### Depth"
            },
            {
              "line": 172,
              "text": "## Lists"
            },
            {
              "line": 174,
              "text": "### When to Use Bullets vs. Numbers"
            },
            {
              "line": 179,
              "text": "### List Formatting"
            },
            {
              "line": 193,
              "text": "### Sentence Fragments OK"
            },
            {
              "line": 203,
              "text": "## Examples"
            },
            {
              "line": 205,
              "text": "### Show, Don't Tell"
            },
            {
              "line": 217,
              "text": "### Minimal, Then Complete"
            },
            {
              "line": 237,
              "text": "### Comments in Code"
            },
            {
              "line": 251,
              "text": "## Error Messages"
            },
            {
              "line": 255,
              "text": "1. Show the exact error message the user sees"
            },
            {
              "line": 268,
              "text": "## Checklist Before Publishing"
            },
            {
              "line": 275,
              "text": "- [ ] Heading structure: H1 \u2192 H2 \u2192 H3 only"
            }
          ]
        }
      ],
      "remote_branch": "test",
      "remote_commit": "2e1961a0850d17d94d855c129183b9d026dd90f7",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/productivity/documentation-docusaurus-aerosim/SKILL.md"
    },
    {
      "skill": "corp-v1-glossary",
      "repository": "home-v1-skills-code-agent/corp-v1-glossary",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "3e84c41320f7a56377251657421f4f9dbcd301396ef2ce2d0059310b718082e4",
          "lines": 77,
          "remote_sha256": "3e84c41320f7a56377251657421f4f9dbcd301396ef2ce2d0059310b718082e4",
          "matches_remote": true,
          "remote_url": "https://gitea.lego-cloud.eu/home-v1-skills-code-agent/corp-v1-glossary/src/commit/24732bf6864a14f2d6aa6cccd7d9fc023438a42d/SKILL.md",
          "excerpts": [
            {
              "line": 5,
              "text": "author: Lego (RootAtSkic), Hermes Agent"
            },
            {
              "line": 14,
              "text": "# Corp v1 Glossary"
            },
            {
              "line": 16,
              "text": "This skill governs project terminology. It keeps term definitions out of architecture, scope, delivery, and other domain pages while preserving links from those pages to one authoritative project Glossary."
            },
            {
              "line": 18,
              "text": "## When to Use"
            },
            {
              "line": 30,
              "text": "## Documentation Contract"
            },
            {
              "line": 37,
              "text": "Term | definition | scope | status | aliases | related terms | authoritative links"
            },
            {
              "line": 45,
              "text": "4. Preserve domain evidence in its owning area. Glossary explains the term; Architecture owns architecture facts and decisions, Scope owns product definitions, and Delivery owns implementation procedures."
            },
            {
              "line": 48,
              "text": "7. Mark uncertain or proposed definitions as `Proposed`; only verified/approved definitions are `Adopted`."
            },
            {
              "line": 51,
              "text": "## Procedure"
            },
            {
              "line": 53,
              "text": "1. Identify the reusable term and the exact sentence that currently explains it. Completion: the candidate is terminology rather than domain evidence or procedure."
            },
            {
              "line": 55,
              "text": "3. Add or update one canonical Glossary entry with scope, status, aliases, related terms, and authoritative links. Completion: one stable term anchor or page owns the definition."
            },
            {
              "line": 56,
              "text": "4. Replace duplicated explanatory prose with a concise contextual statement and Glossary link. Completion: domain pages retain only facts needed for that page."
            },
            {
              "line": 60,
              "text": "## Pitfalls"
            },
            {
              "line": 70,
              "text": "## Verification"
            },
            {
              "line": 75,
              "text": "- [ ] Scope, status, aliases, related terms, and authoritative links are present where applicable."
            },
            {
              "line": 76,
              "text": "- [ ] Domain-owned evidence and decisions remain in their owning sections."
            }
          ]
        }
      ],
      "remote_branch": "test",
      "remote_commit": "24732bf6864a14f2d6aa6cccd7d9fc023438a42d",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/productivity/corp-v1-glossary/SKILL.md"
    },
    {
      "skill": "self-hosted-service-account-recovery",
      "repository": "home-v1-skills-code-agent/self-hosted-service-account-recovery",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "66420934ced92729b5ebd24ab65d323a37938d7a586acff3e17fd6e2b54b9fa1",
          "lines": 73,
          "excerpts": [
            {
              "line": 12,
              "text": "# Self-hosted Service Account Recovery"
            },
            {
              "line": 14,
              "text": "Restore an explicitly authorized automation account on a self-hosted service without borrowing another agent's identity, changing unrelated security settings, or claiming recovery before every access path is independently verified."
            },
            {
              "line": 16,
              "text": "## When to use"
            },
            {
              "line": 25,
              "text": "Do not use it for forgotten passwords, MFA reset, credential rotation, privilege elevation, or another account unless the user explicitly authorizes that exact action."
            },
            {
              "line": 27,
              "text": "## Workflow"
            },
            {
              "line": 29,
              "text": "1. **Freeze identity and scope.** Record the exact service, account, requested target state, and approving human. Account-secret availability is not authorization."
            },
            {
              "line": 30,
              "text": "2. **Prove the failure mode.** Check the account through each affected access path. Distinguish disabled/inactive state from expired credentials, insufficient scope, missing repository permission, proxy failure, or transport failure."
            },
            {
              "line": 31,
              "text": "3. **Discover the supported recovery path.** Prefer the service's authorized administrator UI or API. Inspect the installed version and live command/API contract before choosing a mutation."
            },
            {
              "line": 32,
              "text": "4. **Do not borrow identities implicitly.** Never switch to a sibling agent's token, key, or account merely because it is available. Use another identity only when the user explicitly authorizes it for this recovery."
            },
            {
              "line": 33,
              "text": "5. **Capture pre-state.** Require one exact account record and retain only non-secret fields relevant to recovery."
            },
            {
              "line": 34,
              "text": "6. **Mutate narrowly.** Change only the field that implements the authorized target. Use a conditional/idempotent mutation and require exactly one affected record. Stop on zero or multiple matches."
            },
            {
              "line": 35,
              "text": "7. **Verify the control plane.** Re-read the exact account record in a fresh request/session."
            },
            {
              "line": 36,
              "text": "8. **Verify every original access path.** Reuse the original token/key rather than generating replacements by default. For a source-control account, verify API identity and read-only Git transport independently."
            },
            {
              "line": 37,
              "text": "9. **Resume blocked work only after verification.** Account recovery is complete only when the expected identity and transport succeed; it does not waive task-specific review, CI, approval, or deployment gates."
            },
            {
              "line": 38,
              "text": "10. **Report precisely.** State the account, final active/login state, and independently verified access paths. Never include credentials or shell tokens."
            },
            {
              "line": 40,
              "text": "## Escalation and fallback rules"
            },
            {
              "line": 43,
              "text": "- A direct database fallback requires exact account-level authorization, verified schema/installed version, one-row pre-state, a conditional one-row mutation, and multi-path post-verification."
            },
            {
              "line": 47,
              "text": "## Reference procedures"
            },
            {
              "line": 51,
              "text": "## Pitfalls"
            },
            {
              "line": 62,
              "text": "## Verification checklist"
            },
            {
              "line": 64,
              "text": "- [ ] Exact service account and target state were explicitly authorized."
            },
            {
              "line": 68,
              "text": "- [ ] One exact account record was captured before mutation."
            },
            {
              "line": 69,
              "text": "- [ ] The mutation was conditional and affected exactly one record."
            },
            {
              "line": 72,
              "text": "- [ ] Original secondary transport works independently."
            }
          ]
        },
        {
          "path": "references/gitea-on-truenas-reactivation.md",
          "installed_sha256": "2fab34cb64d5045d7cd8aa906177e4e576ec81c6abe22fa29723284df48e8ded",
          "lines": 50,
          "remote_comparison": "not performed for this supplemental file",
          "excerpts": [
            {
              "line": 1,
              "text": "# Gitea account reactivation on TrueNAS"
            },
            {
              "line": 3,
              "text": "This procedure is a fallback for a Gitea account that is disabled at the database level when the exact account reactivation is authorized and no active Gitea site-admin UI/API identity is available."
            },
            {
              "line": 5,
              "text": "## Preconditions"
            },
            {
              "line": 9,
              "text": "- The user explicitly authorized reactivation of the exact account."
            },
            {
              "line": 12,
              "text": "## Procedure"
            },
            {
              "line": 17,
              "text": "4. Connect through the TrueNAS reverse proxy's `/websocket/shell` route using app-shell options containing the exact app name, current container ID, and bounded command. The middleware-internal `/_shell` route may not be exposed by the reverse proxy."
            },
            {
              "line": 19,
              "text": "6. Prefer a supported admin UI/API/CLI activation operation when present. If absent, use the PostgreSQL container only as a narrow fallback."
            },
            {
              "line": 20,
              "text": "7. Query one exact username and inspect only `name`, `is_active`, and `prohibit_login`. Use PostgreSQL environment variables only in process memory. Do not print their values."
            },
            {
              "line": 22,
              "text": "9. Require exactly one inactive record. Conditionally set only `is_active=true` for that exact username and require `UPDATE 1`. Do not alter password, administrator status, login-prohibition state, tokens, keys, or permissions."
            },
            {
              "line": 25,
              "text": "12. Verify the original SSH key independently with read-only `git ls-remote` against an authorized repository."
            },
            {
              "line": 27,
              "text": "## SQL shape"
            },
            {
              "line": 29,
              "text": "Use parameterization or an equivalently exact identifier binding supported by the execution wrapper:"
            },
            {
              "line": 34,
              "text": "WHERE name = :exact_authorized_username"
            },
            {
              "line": 39,
              "text": "Fail closed unless the result is exactly one row and `UPDATE 1`."
            },
            {
              "line": 41,
              "text": "## Transport and secret handling"
            },
            {
              "line": 45,
              "text": "- Keep API-key authentication on the protected management channel when possible. If the proxied shell requires an internal-LAN plaintext WebSocket because its TLS route is unavailable, transmit only a short-lived single-use shell token\u2014not the long-lived TrueNAS API key."
            },
            {
              "line": 48,
              "text": "## Why independent verification matters"
            },
            {
              "line": 50,
              "text": "The database mutation proves only persisted account state. The Gitea API proves the original token can authenticate as the expected user. SSH proves the separate Git transport and registered key work. All three layers are required before declaring recovery complete."
            }
          ]
        }
      ],
      "remote_readback": "HTTP 404",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/devops/self-hosted-service-account-recovery/SKILL.md"
    },
    {
      "skill": "home-v1--truenas",
      "repository": "home-v1-skills-code-agent/home-v1--truenas",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "33d154a8baf74a3c8072167bfc94223e0fd0df8442709b4bb8318b8e13182d0e",
          "lines": 156,
          "remote_comparison": "not performed for this supplemental file",
          "excerpts": [
            {
              "line": 5,
              "text": "author: RootAtSkic, Hermes Agent"
            },
            {
              "line": 14,
              "text": "# Home v1 TrueNAS"
            },
            {
              "line": 18,
              "text": "## When to Use"
            },
            {
              "line": 26,
              "text": "Do not use this skill to infer live state from old channel messages. Query TrueNAS first when the API is reachable; use preserved evidence only for historical incidents."
            },
            {
              "line": 28,
              "text": "## Sources of Truth"
            },
            {
              "line": 36,
              "text": "When sources disagree, preserve the discrepancy and resolve it before mutation."
            },
            {
              "line": 38,
              "text": "## Prerequisites"
            },
            {
              "line": 40,
              "text": "Use `https://nas.lego-cloud.eu` as the preferred Home v1 TrueNAS management endpoint. Its certificate validates normally, so this hostname should be stored in `HL_V1_TRUENAS_URL` when practical."
            },
            {
              "line": 47,
              "text": "- `HL_V1_TRUENAS_API_KEY` \u2014 API key used only in memory."
            },
            {
              "line": 49,
              "text": "Use `terminal` to check only whether each variable is present. Never print either value, environment dumps, WebSocket request payloads containing authentication data, or exception objects that may include headers or URLs with credentials."
            },
            {
              "line": 53,
              "text": "## How to Run"
            },
            {
              "line": 55,
              "text": "Read-only health summary with the preferred hostname:"
            },
            {
              "line": 67,
              "text": "Read-only method call with the preferred hostname:"
            },
            {
              "line": 87,
              "text": "A method not recognized as read-only is rejected unless `--allow-write` is supplied. That flag is a technical interlock, not authorization."
            },
            {
              "line": 89,
              "text": "## Gondor retained application storage"
            },
            {
              "line": 91,
              "text": "For durable Gondor application data, follow [`references/gondor-retained-application-storage.md`](references/gondor-retained-application-storage.md). The verified pattern is a purpose-specific TrueNAS dataset and NFS share restricted to Gondor node networks, paired with a static Kubernetes PV/PVC using `Retain`. Do not use the default dynamic `truenas-csi` StorageClass for durable databases because its `Delete` reclaim policy can delete the generated dataset when the PVC is removed."
            },
            {
              "line": 93,
              "text": "## Procedure"
            },
            {
              "line": 95,
              "text": "1. **Establish scope.** Identify the exact host, objects, requested outcome, and whether the task is read-only or mutating. Completion criterion: no resource name, method, or authorization boundary is assumed."
            },
            {
              "line": 96,
              "text": "2. **Preflight safely.** Confirm both environment variables are present, normalize the endpoint with `scripts/truenas_rpc.py`, authenticate, and call a harmless method such as `system.info`. Completion criterion: authenticated read succeeds without exposing secrets."
            },
            {
              "line": 97,
              "text": "3. **Discover the live contract.** Inspect `core.get_methods` or matching TrueNAS documentation before using an unfamiliar method, especially on beta or newly upgraded releases. Completion criterion: method name, parameter shape, and job behavior match the live API."
            },
            {
              "line": 98,
              "text": "4. **Capture pre-state.** Query every object that could be affected and retain stable IDs, names, state, and relevant configuration. Completion criterion: a failed or partial mutation can be described exactly."
            },
            {
              "line": 99,
              "text": "5. **Mutate narrowly.** Require explicit user authorization, use the smallest method and parameter set, and pass `--allow-write` only for the approved call. Completion criterion: no unrelated object or setting is changed."
            },
            {
              "line": 100,
              "text": "6. **Handle jobs.** When a call returns a job ID, query the corresponding job until it reaches a terminal state; report errors from the job rather than treating submission as success. Completion criterion: the job is terminal and successful, or its exact failure is known."
            },
            {
              "line": 101,
              "text": "7. **Verify independently.** Re-query the target object with a fresh call and compare the persisted state with the requested outcome. Completion criterion: remote readback\u2014not the mutation response\u2014proves the result."
            },
            {
              "line": 102,
              "text": "8. **Maintain this skill.** Add durable API quirks, safeguards, and reusable logic to `SKILL.md` or `scripts/`; add regression tests; commit, push, and read back the exact remote artifact. Completion criterion: future sessions can reuse the learning without relying on chat history."
            },
            {
              "line": 104,
              "text": "## Safety Rules"
            },
            {
              "line": 106,
              "text": "- Default to read-only discovery. A request to investigate is not permission to change."
            },
            {
              "line": 108,
              "text": "- Never disable TLS verification in normal operation. `--insecure` is diagnostic-only and requires an explicit reason in the report."
            },
            {
              "line": 109,
              "text": "- Never place API keys in Git remotes, command arguments, files, logs, scheduler prompts, or Discord messages."
            },
            {
              "line": 110,
              "text": "- Never run broad mutations, recursive permission changes, pool operations, boot-environment changes, destructive dataset operations, service restarts, or host reboots without exact authorization and a rollback/recovery plan."
            },
            {
              "line": 114,
              "text": "## Script Ownership"
            },
            {
              "line": 116,
              "text": "- `scripts/truenas_rpc.py` owns URL normalization, authentication, JSON-RPC request/response handling, read-only interlocks, timeout behavior, and recursive secret redaction."
            },
            {
              "line": 117,
              "text": "- `scripts/truenas_health.py` owns a compact read-only system/pool/alert summary."
            },
            {
              "line": 122,
              "text": "## Maintenance Contract"
            },
            {
              "line": 131,
              "text": "- a recurring diagnosis can be automated;"
            },
            {
              "line": 136,
              "text": "## Pitfalls"
            },
            {
              "line": 144,
              "text": "- Broad API output may contain credential-shaped fields inside app configuration. Use recursive redaction and select only fields required for the task."
            },
            {
              "line": 146,
              "text": "## Verification"
            },
            {
              "line": 148,
              "text": "Before reporting completion, verify:"
            },
            {
              "line": 150,
              "text": "- the correct channel, host, resource scope, and authorization boundary;"
            },
            {
              "line": 156,
              "text": "- skill changes were committed, pushed, and read back from the private Gitea repository at the exact remote SHA."
            }
          ]
        },
        {
          "path": "references/gondor-retained-application-storage.md",
          "installed_sha256": "2c4c78ef6474872ca808fba47972ea2c54d3e1cf687b3964fb58f2e50f60c4b3",
          "lines": 122,
          "remote_comparison": "not performed for this supplemental file",
          "excerpts": [
            {
              "line": 1,
              "text": "# Gondor retained application storage"
            },
            {
              "line": 5,
              "text": "## Ownership boundary"
            },
            {
              "line": 7,
              "text": "- TrueNAS owns the dataset, filesystem ownership, and NFS export."
            },
            {
              "line": 8,
              "text": "- `home-v1/palantir-v1` owns the idempotent TrueNAS assurance primitives."
            },
            {
              "line": 9,
              "text": "- The application's Gondor template and rendered GitOps repositories own the static PV, PVC, and workload mount."
            },
            {
              "line": 10,
              "text": "- Argo CD consumes the rendered repository only after the TrueNAS prerequisite exists."
            },
            {
              "line": 12,
              "text": "Never rename, edit, re-permission, re-share, or delete an existing dataset for a new application. Create and subsequently manage only the exact application-specific resource authorized by the user."
            },
            {
              "line": 14,
              "text": "## Required naming and isolation"
            },
            {
              "line": 22,
              "text": "For example, an AeroSim integration PostgreSQL volume should use an AeroSim-only name that clearly distinguishes it from production. Do not share a dataset, PVC, PostgreSQL database, role, or credential between production and development/integration testing."
            },
            {
              "line": 24,
              "text": "## Provision in TrueNAS through palantir"
            },
            {
              "line": 46,
              "text": "Before running them:"
            },
            {
              "line": 50,
              "text": "3. Query the exact target dataset and NFS path to prove they are absent or already match."
            },
            {
              "line": 51,
              "text": "4. Determine the workload's real UID/GID and required mode. PostgreSQL images commonly use UID/GID `999`, but verify the selected image rather than assuming it."
            },
            {
              "line": 52,
              "text": "5. Use `maproot=root` when pods run as real non-root UIDs and only client root needs mapping; do not replace this with `mapall` without an explicit ownership reason."
            },
            {
              "line": 53,
              "text": "6. Run dataset and NFS assurance only for the exact authorized resource. The current palantir filesystem wrapper always submits recursive `filesystem.setperm` and polls only the resulting root-path UID; it does not first no-op when ownership already matches. Use that wrapper only on a newly created, verified-empty dataset before application data exists. Never rerun it on populated retained storage. If permissions are wrong after data exists, stop and require exact authorization plus a backup/rollback plan before any permission mutation."
            },
            {
              "line": 58,
              "text": "## Bind it in Gondor GitOps"
            },
            {
              "line": 72,
              "text": "  persistentVolumeReclaimPolicy: Retain"
            },
            {
              "line": 73,
              "text": "  storageClassName: \"\""
            },
            {
              "line": 89,
              "text": "  storageClassName: \"\""
            },
            {
              "line": 96,
              "text": "Use the NFS server and pool/path currently declared by the Gondor template answer file; verify live values before applying. Keep the PV/PVC in a lower Argo CD sync band than the consuming stateful workload when the repository uses ordered applications."
            },
            {
              "line": 98,
              "text": "## Why not the default StorageClass"
            },
            {
              "line": 100,
              "text": "The default `truenas-csi` StorageClass dynamically creates datasets and currently has `reclaimPolicy: Delete`. Reproduce this fact from the live Gondor cluster before each design or mutation:"
            },
            {
              "line": 107,
              "text": "Stop if the class is absent or its observed contract differs from the documentation. With `Delete`, deleting its PVC can delete the generated TrueNAS dataset. Durable databases must use the static retained pattern above unless a separately reviewed retained StorageClass is introduced."
            },
            {
              "line": 109,
              "text": "## Verification"
            },
            {
              "line": 111,
              "text": "Before declaring storage ready, verify all layers independently:"
            },
            {
              "line": 113,
              "text": "1. TrueNAS dataset exists at the exact authorized path."
            },
            {
              "line": 114,
              "text": "2. Filesystem ownership and mode match the verified workload requirements."
            },
            {
              "line": 117,
              "text": "5. The rendered repository contains the same retained objects."
            }
          ]
        }
      ],
      "remote_branch": "test",
      "remote_commit": "f84447bdfcaec5041299c876b03d6369c9f974b9",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/home-v1-skills-code-agent/home-v1--truenas/SKILL.md"
    },
    {
      "skill": "development-durable-wave-execution",
      "repository": "home-v1-skills-code-agent/development-durable-wave-execution",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "c83c784d4a15a1416ddb524bdd6d9bc646f8d55daf46bd5b6ef0a897b7c6cbb8",
          "lines": 121,
          "remote_sha256": "c83c784d4a15a1416ddb524bdd6d9bc646f8d55daf46bd5b6ef0a897b7c6cbb8",
          "matches_remote": true,
          "remote_url": "https://gitea.lego-cloud.eu/home-v1-skills-code-agent/development-durable-wave-execution/src/commit/77416e2ef2c4b2593fcb997653fd6dacc87e24df/SKILL.md",
          "excerpts": [
            {
              "line": 13,
              "text": "# Durable Wave Execution"
            },
            {
              "line": 15,
              "text": "## Purpose"
            },
            {
              "line": 17,
              "text": "Use this skill when an unfinished release or milestone exposes dependency-ordered waves and one worker or recurring job must claim, execute, and finish a complete wave across interruptions. It prevents convenient single-task pickup, duplicate work, abandoned local branches, and scheduler runs that repeatedly observe instead of deliver."
            },
            {
              "line": 21,
              "text": "## Governing handoff"
            },
            {
              "line": 23,
              "text": "Read the workflow's canonical contract before adding gates. When exact task state `READY_FOR_DELIVERY` already means upstream scope, design, implementation approval, allocation, dependency, and readiness checks passed, treat it as the complete pickup handoff. Do not require a second Focus admission, repeat parent approval, or use stale presentation state to strand executable work."
            },
            {
              "line": 27,
              "text": "## Atomic whole-wave claim"
            },
            {
              "line": 33,
              "text": "3. Freeze the complete membership before choosing implementation order."
            },
            {
              "line": 34,
              "text": "4. Claim the **whole wave** as one ownership unit."
            },
            {
              "line": 35,
              "text": "5. Persist every member as active or claimed-and-blocked with one exact blocker."
            },
            {
              "line": 37,
              "text": "7. Start implementation only after the durable claim can be rediscovered by another run."
            },
            {
              "line": 39,
              "text": "Do not claim only the easiest member. A blocked task remains part of the claimed wave; its explicit blocker controls execution order but does not silently release ownership."
            },
            {
              "line": 41,
              "text": "## Continuous execution"
            },
            {
              "line": 43,
              "text": "- Resume all claimed unfinished members before selecting new work."
            },
            {
              "line": 44,
              "text": "- Work in blocker-removal order while preserving one independently reviewable branch per task when practical."
            },
            {
              "line": 46,
              "text": "- After each meaningful transition, persist a checkpoint before continuing."
            },
            {
              "line": 47,
              "text": "- Do not stop at planning, synchronization, or a local-only candidate while safe executable work remains."
            },
            {
              "line": 49,
              "text": "## One-hour closure budget"
            },
            {
              "line": 51,
              "text": "When the owning project sets a one-hour wave target, treat it as an execution constraint rather than an estimate:"
            },
            {
              "line": 53,
              "text": "1. **0\u201310 minutes:** freeze membership, claim every member, resolve dependencies, and start independent workstreams in parallel."
            },
            {
              "line": 55,
              "text": "3. **35\u201350 minutes:** rebase onto the fresh default head, run independent reviews and required broader validation in parallel, then push/open or update PRs immediately."
            },
            {
              "line": 56,
              "text": "4. **50\u201360 minutes:** verify exact-head CI, merge every authorized eligible PR, verify the integration head, and reconcile canonical state."
            },
            {
              "line": 58,
              "text": "At minute 35, stop optional exploration and visual polish. At minute 50, stop repeated local reruns unless a new exact failure requires them. A failed reviewer, missing shell shim, generated test artifact, or stale base must be repaired once and resumed from the last immutable checkpoint; it must not restart the whole wave. Keep Playwright/build outputs outside tracked or untracked repository paths so validation does not invalidate the reviewed packet. At minute 60, report an overrun as a process failure and continue immediately to closure; the only acceptable remaining status is an immutable external blocker with owner and exact evidence, never a generic \u201creview in progress.\u201d"
            },
            {
              "line": 60,
              "text": "## Interruption-safe scheduler contract"
            },
            {
              "line": 62,
              "text": "Recurring-job instructions must be self-contained and explicitly require:"
            },
            {
              "line": 64,
              "text": "1. discovery of existing claims before new selection;"
            },
            {
              "line": 67,
              "text": "4. exact distinction among active, blocked, locally validated, review-pending, pushed, PR-open, merged, and complete states;"
            },
            {
              "line": 70,
              "text": "7. silence only when no claimed work, pickup-ready work, changed blocker, or useful maintenance action exists."
            },
            {
              "line": 72,
              "text": "An in-memory todo, chat statement, or unpushed worktree is not sufficient recovery state. Canonical records carry ownership; remote refs and pull requests carry execution evidence. Reconcile both."
            },
            {
              "line": 74,
              "text": "## Quality gates per task"
            },
            {
              "line": 77,
              "text": "2. Run targeted tests before the full repository validator."
            },
            {
              "line": 79,
              "text": "4. Freeze the complete candidate and obtain the required independent review."
            },
            {
              "line": 80,
              "text": "5. Push and read back the exact branch head."
            },
            {
              "line": 81,
              "text": "6. Verify exact-head CI before merge."
            },
            {
              "line": 82,
              "text": "7. After authorized merge, resolve the integration head independently and verify its CI."
            },
            {
              "line": 85,
              "text": "A transient reviewer, model, runner, or session failure does not release the wave. Keep that task fail-closed at the failed boundary, preserve its exact candidate, continue independent members of the same wave, and retry the gate on the next run. Never relabel local validation as independent review, CI, merge, or completion."
            },
            {
              "line": 87,
              "text": "## Completion condition"
            },
            {
              "line": 89,
              "text": "Wave ownership ends only when every member:"
            },
            {
              "line": 92,
              "text": "- has required review and exact-head CI evidence;"
            },
            {
              "line": 99,
              "text": "## Verification checklist"
            },
            {
              "line": 102,
              "text": "- [ ] Complete wave membership was frozen before implementation ordering."
            },
            {
              "line": 106,
              "text": "- [ ] Replacement runs resume claims before selecting new work."
            },
            {
              "line": 109,
              "text": "- [ ] Wave completion reflects every member, not only successful ones."
            },
            {
              "line": 111,
              "text": "## Pitfalls"
            },
            {
              "line": 114,
              "text": "- Leaving blocked members unclaimed instead of recording exact blockers."
            },
            {
              "line": 117,
              "text": "- Creating branches without persisting canonical ownership."
            },
            {
              "line": 118,
              "text": "- Repeating observation-only scheduled runs while executable work exists."
            }
          ]
        }
      ],
      "remote_branch": "main",
      "remote_commit": "77416e2ef2c4b2593fcb997653fd6dacc87e24df",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/software-development/development-durable-wave-execution/SKILL.md"
    },
    {
      "skill": "development-gates",
      "repository": "home-v1-skills-code-agent/development-gates",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "be7de11ebe0e9d28e796aad0404d28f08db68986e5e9e947c25db52df7e7ed72",
          "lines": 95,
          "remote_sha256": "be7de11ebe0e9d28e796aad0404d28f08db68986e5e9e947c25db52df7e7ed72",
          "matches_remote": true,
          "remote_url": "https://gitea.lego-cloud.eu/home-v1-skills-code-agent/development-gates/src/commit/b5978c45d9e48a255678c6866e7ce12de695e36e/SKILL.md",
          "excerpts": [
            {
              "line": 3,
              "text": "description: \"Use when review, CI, and merge require exact-head proof.\""
            },
            {
              "line": 13,
              "text": "# Development Gates"
            },
            {
              "line": 15,
              "text": "## Purpose"
            },
            {
              "line": 21,
              "text": "## Evidence model"
            },
            {
              "line": 27,
              "text": "- independent `code_review_passed` verdict;"
            },
            {
              "line": 31,
              "text": "- post-merge validation/publication runs."
            },
            {
              "line": 35,
              "text": "## Ordered gates"
            },
            {
              "line": 38,
              "text": "2. Obtain independent logic/security review."
            },
            {
              "line": 39,
              "text": "3. Push only the exact reviewed head and read back its remote ref."
            },
            {
              "line": 40,
              "text": "4. Run every required exact-head remote job."
            },
            {
              "line": 43,
              "text": "7. Resolve the generated integration commit and verify its own workflows."
            },
            {
              "line": 45,
              "text": "If the base advances or conflict resolution changes the tree, synchronize and repeat review. Do not transfer an old-base verdict by task name alone."
            },
            {
              "line": 47,
              "text": "## Authorization continuity"
            },
            {
              "line": 49,
              "text": "Treat authorization as scoped state, not a prompt at every gate. When the user or governing workflow already authorized completing an exact change or approved wave end to end, that authorization remains valid through review, exact-head CI, merge, and required post-merge verification. Do not ask again merely because the work crossed from implementation to merge."
            },
            {
              "line": 51,
              "text": "Reconfirm only when authorization was revoked or narrowed, the candidate changed materially beyond the approved scope, a new destructive/high-impact action appeared, or the owning workflow explicitly reserves that action for a human. Record the authorizing decision and scope with the immutable candidate evidence."
            },
            {
              "line": 53,
              "text": "## Review versus publication readiness"
            },
            {
              "line": 55,
              "text": "A local candidate may be sound before remote CI can exist. Report two decisions:"
            },
            {
              "line": 58,
              "text": "- `publication_ready`: all required remote/runtime evidence also exists."
            },
            {
              "line": 60,
              "text": "Absence of remote CI is not a code defect when the reviewer was explicitly forbidden to push. Such a candidate can be approved for exact-head push while remaining not publication-ready. Conversely, CI success never overrides a deterministic review finding."
            },
            {
              "line": 62,
              "text": "## Attempt integrity"
            },
            {
              "line": 64,
              "text": "A rerun may retain the same Gitea run ID while incrementing `run_attempt`. Treat each attempt as a separate evidence set. Use only the complete job inventory and job IDs from one successful attempt; never splice passing jobs across attempts."
            },
            {
              "line": 66,
              "text": "A transient database or runner incident may justify a fresh full attempt only after confirming that the failure is outside the task delta. Partial focused retries do not satisfy publication gates."
            },
            {
              "line": 68,
              "text": "## Concurrent publication"
            },
            {
              "line": 74,
              "text": "- review and validate that exact artifact;"
            },
            {
              "line": 75,
              "text": "- if it merged prematurely, compare candidate and merge trees and use a forward-only repair;"
            },
            {
              "line": 76,
              "text": "- close unmerged PRs immediately on exact CI failure or deterministic review failure."
            },
            {
              "line": 78,
              "text": "## Verification checklist"
            },
            {
              "line": 88,
              "text": "## Pitfalls"
            },
            {
              "line": 90,
              "text": "- Requiring remote CI while simultaneously forbidding publication."
            }
          ]
        },
        {
          "path": "references/review-ci-merge-contract.md",
          "installed_sha256": "f9b9f18bea1b9ed3e5e3d2f56a311c50cf53655e265f3ed848ff5748605a1bfd",
          "lines": 44,
          "remote_comparison": "not performed for this supplemental file",
          "excerpts": [
            {
              "line": 1,
              "text": "# Review\u2013CI\u2013merge contract"
            },
            {
              "line": 3,
              "text": "## State transitions"
            },
            {
              "line": 7,
              "text": "| Local candidate | clean tree, exact base/head/tree, local gates | independent review |"
            },
            {
              "line": 8,
              "text": "| Review-approved | complete verdict with no logic/security blockers | push exact head |"
            },
            {
              "line": 9,
              "text": "| Remote candidate | remote ref readback equals reviewed head | exact-head CI |"
            },
            {
              "line": 12,
              "text": "| Complete | integration validation and publication gates pass | lifecycle reconciliation |"
            },
            {
              "line": 14,
              "text": "`code_review_passed=true` may coexist with `publication_ready=false` while remote CI is pending. Do not encode an impossible reviewer requirement that remote evidence predate the push that creates it."
            },
            {
              "line": 16,
              "text": "## Authorization scope across gates"
            },
            {
              "line": 18,
              "text": "An exact end-to-end authorization is not consumed by implementation. Unless it is revoked, narrowed, materially exceeded, or the governing workflow explicitly reserves a later action for a human, carry it through review, exact-head CI, merge, and post-merge verification without asking again. Record the authorizing decision and scope; do not infer broader permission from it."
            },
            {
              "line": 20,
              "text": "## Gitea rerun evidence"
            },
            {
              "line": 30,
              "text": "## Diagnosing cross-build failures"
            },
            {
              "line": 32,
              "text": "A host build can pass while a container build fails because Dockerfiles copy only a subset of the repository. For every new compile-time local import, verify all relevant Docker build stages copy it and add a fast static container-contract regression. Still require a Docker-capable exact-head build."
            },
            {
              "line": 36,
              "text": "## Production-path review"
            },
            {
              "line": 38,
              "text": "Search callers from the production entry point. A trace utility, benchmark, browser fixture, test-only adapter, or presentation-only calculation is not shipped integration. Tests must demonstrate an observable effect through the real consumer."
            },
            {
              "line": 40,
              "text": "For memoized initialization, validate every invocation before returning an existing promise and bind reuse to the exact issued configuration. Test cloned, tampered, stale, mixed, and concurrent-different calls."
            },
            {
              "line": 42,
              "text": "## External state mutation"
            }
          ]
        }
      ],
      "remote_branch": "main",
      "remote_commit": "b5978c45d9e48a255678c6866e7ce12de695e36e",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/software-development/development-gates/SKILL.md"
    },
    {
      "skill": "development-integrity",
      "repository": "home-v1-skills-code-agent/development-integrity",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "6082b5198de0ff94b3910e6f398075a155e792076aa2f11eff9bd2c71203cce5",
          "lines": 91,
          "remote_sha256": "6082b5198de0ff94b3910e6f398075a155e792076aa2f11eff9bd2c71203cce5",
          "matches_remote": true,
          "remote_url": "https://gitea.lego-cloud.eu/home-v1-skills-code-agent/development-integrity/src/commit/b677c99f24cbe83734a9d6a495e388837fc88701/SKILL.md",
          "excerpts": [
            {
              "line": 8,
              "text": "    tags: [delivery, status, ci, pull-requests, escalation, closure]"
            },
            {
              "line": 12,
              "text": "# Development Integrity"
            },
            {
              "line": 14,
              "text": "## Purpose"
            },
            {
              "line": 18,
              "text": "## Evidence-first status sequence"
            },
            {
              "line": 20,
              "text": "Before reporting status, verify the current remote checkpoint in this order:"
            },
            {
              "line": 22,
              "text": "1. local worktree and exact candidate identity;"
            },
            {
              "line": 24,
              "text": "3. required pull request exists and points at that exact head;"
            },
            {
              "line": 25,
              "text": "4. exact-head candidate CI exists and has the stated result;"
            },
            {
              "line": 26,
              "text": "5. required independent review covers that same exact head;"
            },
            {
              "line": 27,
              "text": "6. merge state and exact integration head;"
            },
            {
              "line": 28,
              "text": "7. integration CI for that exact merge head;"
            },
            {
              "line": 31,
              "text": "Never report \u201cwaiting for CI\u201d unless the required PR exists and an exact-head run is actually queued or running. Inspect workflow triggers first. If CI starts only after merge, say that candidate CI is not running and continue through the authorized review/merge checkpoint."
            },
            {
              "line": 35,
              "text": "## Plain-language reporting"
            },
            {
              "line": 37,
              "text": "Lead with task, wave, and verified status. Keep the default report short and nontechnical. Add one blocker and one next action only when useful. Provide SHAs, run IDs, changed files, and detailed logs only when requested or required for an audit."
            },
            {
              "line": 43,
              "text": "- **Implementation completion:** accepted tasks integrated with required exact-head and integration gates."
            },
            {
              "line": 48,
              "text": "## Continuous-delivery guardrails"
            },
            {
              "line": 52,
              "text": "- If a PR merges while a new finding is being fixed, create a forward-fix PR immediately and verify its own review, CI, merge, and integration CI."
            },
            {
              "line": 53,
              "text": "- Treat a reminder or buzzer cron as a notifier, never as an executor."
            },
            {
              "line": 54,
              "text": "- Do not pause independent wave members because one task is blocked unless dependency evidence requires it."
            },
            {
              "line": 56,
              "text": "## Ask for help early"
            },
            {
              "line": 58,
              "text": "After one focused repair attempt or one hour without a meaningful checkpoint, whichever comes first:"
            },
            {
              "line": 60,
              "text": "1. state the exact blocker in plain language;"
            },
            {
              "line": 61,
              "text": "2. ask the relevant team or domain owner for help;"
            },
            {
              "line": 62,
              "text": "3. assign an independent repair/review worker when possible; and"
            },
            {
              "line": 63,
              "text": "4. continue independent work in parallel."
            },
            {
              "line": 65,
              "text": "Repeated solo retries are not progress. A status after the threshold must name the helper, owner, or external dependency\u2014not only \u201cin progress.\u201d"
            },
            {
              "line": 67,
              "text": "## Completion test"
            },
            {
              "line": 69,
              "text": "A task is delivered only when its required candidate review/CI, integration, and integration CI are verified against exact identities. A wave is formally closed only when every member is delivered, canonical lifecycle records are reconciled, and the release handoff is recorded. Old superseded branches or failed PR attempts are historical evidence, not open work, when a later verified PR fully replaced them."
            },
            {
              "line": 71,
              "text": "## Pitfalls"
            },
            {
              "line": 76,
              "text": "- Treating a merged task as formally closed before lifecycle reconciliation."
            },
            {
              "line": 77,
              "text": "- Letting a status cron substitute for an executing agent."
            },
            {
              "line": 80,
              "text": "- Reporting all wave members as active when only one has an executor."
            },
            {
              "line": 82,
              "text": "## Verification checklist"
            },
            {
              "line": 85,
              "text": "- [ ] PR existence, exact head, and mergeability were checked."
            },
            {
              "line": 87,
              "text": "- [ ] Independent review covers the exact candidate being merged."
            },
            {
              "line": 90,
              "text": "- [ ] Blockers beyond the threshold were escalated to a named helper or owner."
            }
          ]
        },
        {
          "path": "references/remote-checkpoint-matrix.md",
          "installed_sha256": "8a14fbc9b5c530e303a22e219105581c64b7a670f8e622e5d2e536cdce99f7fe",
          "lines": 27,
          "remote_comparison": "not performed for this supplemental file",
          "excerpts": [
            {
              "line": 1,
              "text": "# Remote checkpoint matrix"
            },
            {
              "line": 5,
              "text": "| Local candidate | Clean worktree and exact local commit |"
            },
            {
              "line": 6,
              "text": "| Pushed | Remote branch equals the exact local commit |"
            },
            {
              "line": 7,
              "text": "| PR open | PR readback shows the expected source, base, and exact head |"
            },
            {
              "line": 8,
              "text": "| CI running | Exact-head run exists and is queued or running |"
            },
            {
              "line": 9,
              "text": "| Candidate green | Every required exact-head job is terminal success |"
            },
            {
              "line": 10,
              "text": "| Reviewed | Independent review names the same exact head/tree and has no blockers |"
            },
            {
              "line": 12,
              "text": "| Integration green | Required integration jobs passed at the exact merge commit |"
            },
            {
              "line": 16,
              "text": "## Status wording"
            },
            {
              "line": 19,
              "text": "- Say **fixing** only when a named worker/process is actively changing the candidate."
            },
            {
              "line": 20,
              "text": "- Say **waiting for CI** only when an exact run is visible."
            },
            {
              "line": 21,
              "text": "- Say **blocked** with a concrete cause, owner/helper, and next action."
            },
            {
              "line": 23,
              "text": "- Say **closed** only after both delivery and formal closure evidence exist."
            },
            {
              "line": 25,
              "text": "## Concurrency checks"
            },
            {
              "line": 27,
              "text": "Immediately before merge, re-read the PR head, review identity, CI, base, and mergeability. After merge, independently resolve the integration head and its CI. If the head or base changes, invalidate the old review/CI claim and establish a fresh candidate. If another actor merges while a new repair is pending, open a forward-fix PR rather than treating the old merge as current acceptance."
            }
          ]
        }
      ],
      "remote_branch": "main",
      "remote_commit": "b677c99f24cbe83734a9d6a495e388837fc88701",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/software-development/development-integrity/SKILL.md"
    },
    {
      "skill": "development-method-ttd",
      "repository": "home-v1-skills-code-agent/development-method-ttd",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "02fd47dbbcabaefbee9881be20cb171a3c6c674e6c30056c466f65769f723754",
          "lines": 362,
          "remote_sha256": "02fd47dbbcabaefbee9881be20cb171a3c6c674e6c30056c466f65769f723754",
          "matches_remote": true,
          "remote_url": "https://gitea.lego-cloud.eu/home-v1-skills-code-agent/development-method-ttd/src/commit/2155b0a250cea282bf6628231d8e3de286522a86/SKILL.md",
          "excerpts": [
            {
              "line": 3,
              "text": "description: \"TDD: enforce RED-GREEN-REFACTOR, tests before code.\""
            },
            {
              "line": 14,
              "text": "# Development Method TTD"
            },
            {
              "line": 16,
              "text": "## Overview"
            },
            {
              "line": 24,
              "text": "## When to Use"
            },
            {
              "line": 39,
              "text": "## The Iron Law"
            },
            {
              "line": 42,
              "text": "NO PRODUCTION CODE WITHOUT A FAILING TEST FIRST"
            },
            {
              "line": 45,
              "text": "Write code before the test? Delete it. Start over."
            },
            {
              "line": 55,
              "text": "## Red-Green-Refactor Cycle"
            },
            {
              "line": 57,
              "text": "### RED \u2014 Write Failing Test"
            },
            {
              "line": 95,
              "text": "### Verify RED \u2014 Watch It Fail"
            },
            {
              "line": 100,
              "text": "# Use terminal tool to run the specific test"
            },
            {
              "line": 113,
              "text": "### GREEN \u2014 Minimal Code"
            },
            {
              "line": 141,
              "text": "### Verify GREEN \u2014 Watch It Pass"
            },
            {
              "line": 146,
              "text": "# Run the specific test"
            },
            {
              "line": 149,
              "text": "# Then run ALL tests to check for regressions"
            },
            {
              "line": 162,
              "text": "### REFACTOR \u2014 Clean Up"
            },
            {
              "line": 164,
              "text": "After green only:"
            },
            {
              "line": 174,
              "text": "### Repeat"
            },
            {
              "line": 178,
              "text": "## Avoid Horizontal Slices"
            },
            {
              "line": 180,
              "text": "Do **not** write all tests first and then all implementation. That is horizontal slicing: RED becomes \"write a pile of imagined tests\" and GREEN becomes \"make the pile pass.\" It produces brittle tests because the tests are designed before the implementation has taught you what behavior and interface actually matter."
            },
            {
              "line": 197,
              "text": "## Why Order Matters"
            },
            {
              "line": 209,
              "text": "**\"I already manually tested all the edge cases\"**"
            },
            {
              "line": 221,
              "text": "Sunk cost fallacy. The time is already gone. Your choice now:"
            },
            {
              "line": 230,
              "text": "- Finds bugs before commit (faster than debugging after)"
            },
            {
              "line": 241,
              "text": "Tests-after are biased by your implementation. You test what you built, not what's required. Tests-first force edge case discovery before implementing."
            },
            {
              "line": 243,
              "text": "## Common Rationalizations"
            },
            {
              "line": 250,
              "text": "| \"Already manually tested\" | Ad-hoc \u2260 systematic. No record, can't re-run. |"
            },
            {
              "line": 255,
              "text": "| \"TDD will slow me down\" | TDD faster than debugging. Pragmatic = test-first. |"
            },
            {
              "line": 259,
              "text": "## Red Flags \u2014 STOP and Start Over"
            },
            {
              "line": 263,
              "text": "- Code before test"
            },
            {
              "line": 269,
              "text": "- \"I already manually tested it\""
            },
            {
              "line": 272,
              "text": "- \"Already spent X hours, deleting is wasteful\""
            },
            {
              "line": 278,
              "text": "## Verification Checklist"
            },
            {
              "line": 280,
              "text": "Before marking work complete:"
            },
            {
              "line": 283,
              "text": "- [ ] Watched each test fail before implementing"
            },
            {
              "line": 288,
              "text": "- [ ] Tests use real code (mocks only if unavoidable)"
            },
            {
              "line": 293,
              "text": "## When Stuck"
            },
            {
              "line": 302,
              "text": "## Hermes Agent Integration"
            },
            {
              "line": 304,
              "text": "### Running Tests"
            },
            {
              "line": 309,
              "text": "# RED \u2014 verify failure"
            },
            {
              "line": 312,
              "text": "# GREEN \u2014 verify pass"
            },
            {
              "line": 315,
              "text": "# Full suite \u2014 verify no regressions"
            },
            {
              "line": 319,
              "text": "### With delegate_task"
            },
            {
              "line": 342,
              "text": "### With systematic-debugging"
            },
            {
              "line": 348,
              "text": "## Testing Anti-Patterns"
            },
            {
              "line": 352,
              "text": "- **Happy path only** \u2014 always test edge cases, errors, and boundaries"
            },
            {
              "line": 355,
              "text": "## Final Rule"
            }
          ]
        }
      ],
      "remote_branch": "main",
      "remote_commit": "2155b0a250cea282bf6628231d8e3de286522a86",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/software-development/development-method-ttd/SKILL.md"
    },
    {
      "skill": "development-branching-strategy-aerosim",
      "repository": "corp-v1-aerosim-skills-code-agent/development-branching-strategy-aerosim",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "938dbccff02604104f27c611138558659f8872aeb6d837c17fc1dc26d6b86cbc",
          "lines": 115,
          "remote_sha256": "184b5c2cbf71c77e66f77e6353989ae5894ee011a43de4e042cf8d6e6d6aca00",
          "matches_remote": false,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/development-branching-strategy-aerosim/src/commit/ff4026cb190e35242412849d97bc9d5b172e7402/SKILL.md",
          "excerpts": [
            {
              "line": 7,
              "text": "  or merging in any repository \u2014 even if the user only says \"new branch\", \"commit this\","
            },
            {
              "line": 9,
              "text": "  X.Y.Z.Q/{TASK_MANAGEMENT_CODE}-{FEATURE_X}-{SEQ} with the version read from git tags"
            },
            {
              "line": 10,
              "text": "  only. Also use when deciding what an agent may do autonomously in a repository versus"
            },
            {
              "line": 21,
              "text": "# Branching Strategy"
            },
            {
              "line": 24,
              "text": "## Standing AeroSim PR merge authority"
            },
            {
              "line": 26,
              "text": "RootAtSkic authorizes Hermes to merge PRs across all AeroSim repositories and channels without per-PR human authorization. Source: [explicit standing delegation](https://discord.com/channels/1518726359512387766/1537544173366943886/1554873863165182015)."
            },
            {
              "line": 28,
              "text": "This project-local rule supersedes older instructions in this skill, linked references, or inherited guidance requiring a human to authorize PR merging or forbidding Hermes from merging. Preserve it during skill adoption and synchronization. Complete required review and exact-head CI, freshly verify the unchanged mergeable head, merge it, and verify the integration revision and required post-merge checks. Do not block a reviewed, green PR for owner merge approval or request repeated permission."
            },
            {
              "line": 30,
              "text": "This delegates PR integration, not product/design/architecture decisions, human UAT acceptance, or independently gated release/deployment actions. If merging a GitOps or release PR itself triggers a separately gated deployment, establish that deployment authority first; do not relabel it as a PR-merge approval. Repository protections, security checks, and genuine technical blockers still apply."
            },
            {
              "line": 36,
              "text": "## Main rules"
            },
            {
              "line": 38,
              "text": "### Rule 1 \u2014 All local work happens on a branch created from `origin/test`"
            },
            {
              "line": 46,
              "text": "- **Before creating the branch, present the computed branch name to the user and get"
            },
            {
              "line": 47,
              "text": "  their confirmation** \u2014 run the `next` action, show the name, create only after a yes."
            },
            {
              "line": 49,
              "text": "### Rule 2 \u2014 Branch naming convention"
            },
            {
              "line": 55,
              "text": "Example: `4.0.0.0/IIAA-123-api-cleanup-001`"
            },
            {
              "line": 59,
              "text": "| `X.Y.Z.Q` | Version | **Git tags only** \u2014 latest semver tag of the target repo (`v` prefix allowed; 3-part tags get `Q=0`). No tags \u2192 stop and ask which version to tag, never guess. |"
            },
            {
              "line": 62,
              "text": "| `SEQ` | 3-digit sequence, `001`-based | Auto-incremented by scanning existing local + remote branches with the same prefix |"
            },
            {
              "line": 66,
              "text": "### Rule 3 \u2014 Merging into `test` is via Pull Request, by the user, always"
            },
            {
              "line": 73,
              "text": "## Supporting principles"
            },
            {
              "line": 75,
              "text": "- **Commit** only when the user asks; **push** only when the user asks, and only branches."
            },
            {
              "line": 81,
              "text": "- Versions are marked with git tags \u2014 tags are the single source of truth."
            },
            {
              "line": 83,
              "text": "## Available scripts"
            },
            {
              "line": 88,
              "text": "# Compute the next branch name + detected base (no changes made) \u2014 show this to the user"
            },
            {
              "line": 91,
              "text": "# After user confirmation: fetch, then create the branch from origin/test (or origin/main)"
            },
            {
              "line": 94,
              "text": "# Build TypeScript to dist/"
            },
            {
              "line": 102,
              "text": "## Workflow"
            },
            {
              "line": 107,
              "text": "4. Run `task create` only after the user confirms. If it reports no `test`/`main` on the remote, ask the user and re-run with `--base`."
            },
            {
              "line": 110,
              "text": "## Gotchas"
            }
          ]
        }
      ],
      "remote_branch": "test",
      "remote_commit": "ff4026cb190e35242412849d97bc9d5b172e7402",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/software-development/development-branching-strategy-aerosim/SKILL.md"
    },
    {
      "skill": "development-monorepo-pnpm",
      "repository": "home-v1-skills-code-agent/development-monorepo-pnpm",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "267aa0dee0836760cbe689fb461e3e7d618f512ab04898441f5b007fdc8bc6ba",
          "lines": 314,
          "remote_sha256": "267aa0dee0836760cbe689fb461e3e7d618f512ab04898441f5b007fdc8bc6ba",
          "matches_remote": true,
          "remote_url": "https://gitea.lego-cloud.eu/home-v1-skills-code-agent/development-monorepo-pnpm/src/commit/101de8353f6784e6c36282e6f215e4a274a7f898/SKILL.md",
          "excerpts": [
            {
              "line": 11,
              "text": "  cutting changesets; editing Dockerfiles, Helm charts, or Turborepo tasks; running filtered"
            },
            {
              "line": 13,
              "text": "  build-pipeline problems \u2014 even when the user only names their repository instead of saying"
            },
            {
              "line": 23,
              "text": "# Full-Stack pnpm + Turborepo Monorepo"
            },
            {
              "line": 32,
              "text": "## Scope routing"
            },
            {
              "line": 42,
              "text": "| Dockerfiles, images, Helm charts, values, deploy automation | `references/containers-and-helm.md` |"
            },
            {
              "line": 44,
              "text": "## 1. Canonical repository shape"
            },
            {
              "line": 56,
              "text": "\u251c\u2500\u2500 helm-charts/<chart-name>/      # deployment assets"
            },
            {
              "line": 57,
              "text": "\u251c\u2500\u2500 .scripts/                      # shell modules: docker, images, helm, pipelines"
            },
            {
              "line": 63,
              "text": "\u2514\u2500\u2500 Taskfile.yml                   # thin delegator \u2014 includes .scripts/ modules only"
            },
            {
              "line": 67,
              "text": "# pnpm-workspace.yaml"
            },
            {
              "line": 76,
              "text": "`applications-backend/<name>/core`. The four-way split already gives each side its own"
            },
            {
              "line": 79,
              "text": "**A workspace exists only if a glob matches it.** Creating a directory is not enough: confirm"
            },
            {
              "line": 83,
              "text": "### Why the split is in the directory name"
            },
            {
              "line": 88,
              "text": "When both live under one `packages/` root, the only thing standing between a secret-handling"
            },
            {
              "line": 91,
              "text": "## 2. Default stacks"
            },
            {
              "line": 100,
              "text": "*before* scaffolding. If the architecture is silent, use the defaults above and record the"
            },
            {
              "line": 107,
              "text": "## 3. The one sanctioned cross-boundary dependency"
            },
            {
              "line": 110,
              "text": "import from `packages-frontend/*` \u2014 with exactly one exception:"
            },
            {
              "line": 112,
              "text": "**API contract types.** The service that defines an endpoint owns its contract, so contract"
            },
            {
              "line": 113,
              "text": "packages live under `packages-backend/<domain>-contracts` and expose **types only** \u2014 no"
            },
            {
              "line": 114,
              "text": "runtime code, no dependencies, no side effects. Frontend consumers import them type-only:"
            },
            {
              "line": 117,
              "text": "import type { UserDto } from \"@<scope>/user-contracts\";"
            },
            {
              "line": 125,
              "text": "Split it before it ships anything into a bundle."
            },
            {
              "line": 127,
              "text": "## 4. Tooling baseline"
            },
            {
              "line": 130,
              "text": "(exact pnpm version, read by corepack), `engines.node`, `pnpm.overrides` for security pins,"
            },
            {
              "line": 140,
              "text": "pnpm prepush         # turbo build --filter=...[HEAD^]  \u2014 only what the branch touched"
            },
            {
              "line": 144,
              "text": "build failure for every downstream workspace, not a warning. Full pipeline, cache outputs, and"
            },
            {
              "line": 147,
              "text": "## 5. Before making changes"
            },
            {
              "line": 149,
              "text": "1. Identify the exact workspace path, and therefore which of the four roots you are in."
            },
            {
              "line": 151,
              "text": "3. Inspect its `tsconfig.json`, lint config, build config, and source layout before adding."
            },
            {
              "line": 153,
              "text": "5. For Module Federation changes, map the host\u2194remote contract before editing either side."
            },
            {
              "line": 164,
              "text": "## 6. Implementation rules"
            },
            {
              "line": 166,
              "text": "### Workspace boundaries"
            },
            {
              "line": 168,
              "text": "- Application logic stays inside its own workspace."
            },
            {
              "line": 169,
              "text": "- Promote into a shared package only when **more than one** application needs it *and* the"
            },
            {
              "line": 176,
              "text": "### TypeScript"
            },
            {
              "line": 184,
              "text": "### Dependencies"
            },
            {
              "line": 187,
              "text": "- Root dependencies are monorepo tooling only \u2014 turbo, prettier, husky, changesets,"
            },
            {
              "line": 195,
              "text": "### Formatting"
            },
            {
              "line": 200,
              "text": "### Playwright Chromium rendering in GPU-enabled CI"
            },
            {
              "line": 223,
              "text": "Treat screenshot differences as a possible renderer-baseline change before classifying them"
            },
            {
              "line": 227,
              "text": "## 7. Validation"
            },
            {
              "line": 229,
              "text": "Validate the narrowest scope covering the change, then widen if it touched shared code."
            },
            {
              "line": 232,
              "text": "# Single workspace"
            },
            {
              "line": 233,
              "text": "pnpm check-types --filter=\"@<scope>/<name>\""
            },
            {
              "line": 234,
              "text": "pnpm lint        --filter=\"@<scope>/<name>\""
            },
            {
              "line": 235,
              "text": "pnpm build       --filter=\"@<scope>/<name>\""
            },
            {
              "line": 237,
              "text": "# Shared package or cross-workspace change \u2014 unfiltered, plus each consumer"
            },
            {
              "line": 239,
              "text": "pnpm build --filter=\"@<scope>/<consumer>\""
            },
            {
              "line": 241,
              "text": "# Only what the branch changed"
            },
            {
              "line": 249,
              "text": "pnpm dev --filter=\"@<scope>/<host>\""
            },
            {
              "line": 250,
              "text": "pnpm dev --filter=\"@<scope>/<remote>\""
            },
            {
              "line": 256,
              "text": "## 8. Blast-radius checks"
            },
            {
              "line": 258,
              "text": "Before changing anything another workspace can observe \u2014 an exported symbol, a REST base path,"
            },
            {
              "line": 265,
              "text": "5. Validate each affected consumer with its own `--filter` build."
            },
            {
              "line": 268,
              "text": "grep -Rn \"from '@<scope>/<package>'\" applications-backend applications-frontend \\"
            },
            {
              "line": 273,
              "text": "A port or base path change is never local: it also lands in the Helm chart values, the gateway"
            },
            {
              "line": 274,
              "text": "routing, and every frontend caller. See `references/containers-and-helm.md`."
            },
            {
              "line": 276,
              "text": "## 9. Durable records"
            },
            {
              "line": 285,
              "text": "## 10. Common mistakes to avoid"
            },
            {
              "line": 292,
              "text": "- Changing a service port or base path without updating Helm values, gateway routing, and"
            },
            {
              "line": 294,
              "text": "- Changing an MFE `exposes` key, remote name, or remote URL on one side only."
            },
            {
              "line": 297,
              "text": "- Skipping `--filter` validation for the exact workspace you edited."
            },
            {
              "line": 305,
              "text": "## Reference files"
            },
            {
              "line": 313,
              "text": "| `references/containers-and-helm.md` | `turbo prune` image builds for both sides, chart layout, values and secrets, port/base-path propagation | Touching Dockerfiles, images, or charts |"
            }
          ]
        }
      ],
      "remote_branch": "test",
      "remote_commit": "101de8353f6784e6c36282e6f215e4a274a7f898",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/software-development/development-monorepo-pnpm/SKILL.md"
    },
    {
      "skill": "development-scripts-aerosim",
      "repository": "corp-v1-aerosim-skills-code-agent/development-scripts-aerosim",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "80c2c7c51f9c031e7a31c41d99d3ef6ecf088e76fd495334b92f3f77b80425e3",
          "lines": 211,
          "remote_sha256": "191566bac10096536b724069dac0ab6b93bb46977cab3635ae3b7e5f0afd86b2",
          "matches_remote": false,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/development-scripts-aerosim/src/commit/80b6f9ec93798995ba187a6c410c23b4ba26c47d/SKILL.md",
          "excerpts": [
            {
              "line": 12,
              "text": "  order, and secret-safety \u2014 even when the user only says \".scripts\", \"Taskfile\", \"task"
            },
            {
              "line": 19,
              "text": "compatibility: Requires bash. Taskfile (task) needed only for task-runner integration. Designed for Cline, Claude Code, GitHub Copilot, OpenAI Codex, and other compatible agent environments"
            },
            {
              "line": 22,
              "text": "# Repository Script Modules"
            },
            {
              "line": 25,
              "text": "## Standing AeroSim PR merge authority"
            },
            {
              "line": 27,
              "text": "RootAtSkic authorizes Hermes to merge PRs across all AeroSim repositories and channels without per-PR human authorization. Source: [explicit standing delegation](https://discord.com/channels/1518726359512387766/1537544173366943886/1554873863165182015)."
            },
            {
              "line": 29,
              "text": "This project-local rule supersedes older instructions in this skill, linked references, or inherited guidance requiring a human to authorize PR merging or forbidding Hermes from merging. Preserve it during skill adoption and synchronization. Complete required review and exact-head CI, freshly verify the unchanged mergeable head, merge it, and verify the integration revision and required post-merge checks. Do not block a reviewed, green PR for owner merge approval or request repeated permission."
            },
            {
              "line": 31,
              "text": "This delegates PR integration, not product/design/architecture decisions, human UAT acceptance, or independently gated release/deployment actions. If merging a GitOps or release PR itself triggers a separately gated deployment, establish that deployment authority first; do not relabel it as a PR-merge approval. Repository protections, security checks, and genuine technical blockers still apply."
            },
            {
              "line": 42,
              "text": "## Scope routing"
            },
            {
              "line": 52,
              "text": "| Creating a module, changing one, or checking work before delivery | `references/workflows.md` |"
            },
            {
              "line": 58,
              "text": "## 1. Repository model"
            },
            {
              "line": 67,
              "text": "      Taskfile.yml              # only when the module is exposed through task"
            },
            {
              "line": 78,
              "text": "(`. ./.scripts/loggers/lib/--index-api.sh`). This is why running a script from inside its own"
            },
            {
              "line": 82,
              "text": "## 2. Module anatomy"
            },
            {
              "line": 105,
              "text": "| `Taskfile.yml` | Maps task names to `api/` scripts, forwarding `{{.CLI_ARGS}}`. |"
            },
            {
              "line": 107,
              "text": "Not every module has every file \u2014 a support-only library like `loggers` has no `api/` and no"
            },
            {
              "line": 110,
              "text": "## 3. Standard runtime flow"
            },
            {
              "line": 123,
              "text": "**The order is load-bearing.** The reader runs before the validator so the validator has"
            },
            {
              "line": 124,
              "text": "something to check; both run before the API index so implementation functions see a"
            },
            {
              "line": 127,
              "text": "Arguments reach the reader because sourcing happens with the wrapper's own arguments still in"
            },
            {
              "line": 128,
              "text": "scope \u2014 this is why `api/` wrappers need no argument forwarding of their own."
            },
            {
              "line": 130,
              "text": "## 4. Naming conventions"
            },
            {
              "line": 152,
              "text": "## 5. Non-negotiable rules"
            },
            {
              "line": 156,
              "text": "- **CLI flags are parsed only in `--env-vars-reader.sh`.** Never in an implementation file."
            },
            {
              "line": 159,
              "text": "- **Readers preserve already-exported values** \u2014 `export X=\"${X:-default}\"` \u2014 so a caller can"
            },
            {
              "line": 160,
              "text": "  override any setting before invoking the module. A module reader never sources `.env` files"
            },
            {
              "line": 161,
              "text": "  itself; the `base` module owns that. See `references/environment-model.md`."
            },
            {
              "line": 169,
              "text": "## 6. Creating a module"
            },
            {
              "line": 177,
              "text": "4. Add `api/<command>.sh` wrappers for user-facing commands only."
            },
            {
              "line": 178,
              "text": "5. Add a module `Taskfile.yml`, and a root include only if the namespace is stable."
            },
            {
              "line": 181,
              "text": "## 7. Validation"
            },
            {
              "line": 184,
              "text": "# Syntax-check every script in the module"
            },
            {
              "line": 187,
              "text": "# Confirm task exposure resolves"
            },
            {
              "line": 190,
              "text": "# Run the command itself, from the repository root"
            },
            {
              "line": 194,
              "text": "`bash -n` catches syntax errors but **not** a missing entry in `--index-api.sh` \u2014 that only"
            },
            {
              "line": 195,
              "text": "surfaces at call time. Always run the command once, from the repository root, before"
            },
            {
              "line": 198,
              "text": "## 8. Mistakes to avoid"
            },
            {
              "line": 204,
              "text": "  environment-specific files as a side effect. Use `base/lib/--index-api.sh` when you only"
            }
          ]
        }
      ],
      "remote_branch": "test",
      "remote_commit": "80b6f9ec93798995ba187a6c410c23b4ba26c47d",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/software-development/development-scripts-aerosim/SKILL.md"
    },
    {
      "skill": "devsecops-ci-cd-gitea",
      "repository": "home-v1-skills-code-agent/devsecops-ci-cd-gitea",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "6ff2d728d547a0b2191106cb7c6296c648f5a0d84f81a64617eb37a089df58f5",
          "lines": 144,
          "remote_sha256": "6ff2d728d547a0b2191106cb7c6296c648f5a0d84f81a64617eb37a089df58f5",
          "matches_remote": true,
          "remote_url": "https://gitea.lego-cloud.eu/home-v1-skills-code-agent/devsecops-ci-cd-gitea/src/commit/be07b8183518b20e8302bca06dafe436285803cd/SKILL.md",
          "excerpts": [
            {
              "line": 5,
              "text": "  Helm charts for every application in a pnpm + Turborepo monorepo, pushing both to a Harbor"
            },
            {
              "line": 8,
              "text": "  `applications-frontend/<app>/Dockerfile`; packaging and pushing OCI Helm charts from"
            },
            {
              "line": 9,
              "text": "  `helm-charts/<chart>/`; wiring `docker login` / `helm registry login` against Harbor with a"
            },
            {
              "line": 12,
              "text": "  user only says \"the CI/CD thing\", \"the docker pipeline\", or names their monorepo."
            },
            {
              "line": 21,
              "text": "  a runner whose jobs can run docker and helm. Designed for Claude Code, Cline, GitHub Copilot,"
            },
            {
              "line": 25,
              "text": "# DevSecOps CI/CD \u2014 Gitea Actions \u2192 Harbor"
            },
            {
              "line": 29,
              "text": "`helm-charts/`, package and push it as an OCI artifact. Both land in **Harbor**, authenticated"
            },
            {
              "line": 32,
              "text": "This skill owns the pipeline up to *\"the artifact is in Harbor\"*. Deploying that artifact to"
            },
            {
              "line": 36,
              "text": "## When to use \u2014 and which reference to open"
            },
            {
              "line": 40,
              "text": "| Wire or fix registry login (`docker login` / `helm registry login`), robot accounts, or Gitea secret/variable names | `references/harbor-auth.md` |"
            },
            {
              "line": 43,
              "text": "| Prove a Playwright job container exposes the render device, compare software and GPU Chromium, or collect exact-SHA renderer/timing evidence | `references/playwright-gpu-actions.md` |"
            },
            {
              "line": 44,
              "text": "| Design secure branch/PR/default-branch separation, immutable retry, and publication evidence | `references/pipeline-workflow.md` |"
            },
            {
              "line": 46,
              "text": "## The mental model"
            },
            {
              "line": 52,
              "text": "                          \u2514\u2500 helm-charts/<chart>/Chart.yaml         \u2500\u25b6 helm package            \u2500\u25b6 push  oci://harbor/<project>/charts"
            },
            {
              "line": 58,
              "text": "## 1. Registry authentication (the \"docker thing\")"
            },
            {
              "line": 60,
              "text": "Harbor push/pull uses a **project-scoped robot account** \u2014 a non-human account with a long"
            },
            {
              "line": 67,
              "text": "| Kind | Name | Scope in Gitea | Example value |"
            },
            {
              "line": 74,
              "text": "Why this split: the username, address, and project are **not secret**, so they go in variables. The robot secret is the only sensitive value and remains an organization secret because Gitea has no global-secret level."
            },
            {
              "line": 80,
              "text": "echo \"${HARBOR_SECRET}\" | helm  registry login \"${HARBOR_ADDRESS}\" -u \"${HARBOR_USERNAME}\" --password-stdin"
            },
            {
              "line": 83,
              "text": "## 2. What gets built, and from where"
            },
            {
              "line": 92,
              "text": "- **Charts**: one per `helm-charts/*/Chart.yaml`. Charts are **centralized under `helm-charts/`"
            },
            {
              "line": 94,
              "text": "  `applications-frontend/<app>/helm` \u2014 that is not where charts live in this monorepo family."
            },
            {
              "line": 95,
              "text": "- **Identity**: use the full 40-character source revision for governed branch publication. Do not publish shortened SHAs or a mutable `latest` alias when GitOps consumes immutable artifacts. See `references/pipeline-workflow.md` \u00a7Immutable identity and retry."
            },
            {
              "line": 97,
              "text": "## 3. Publish gating \u2014 build always, push selectively"
            },
            {
              "line": 101,
              "text": "| pull request / feature branch | build only (verify it compiles) | `helm lint` + `helm package` (verify) |"
            },
            {
              "line": 102,
              "text": "| push to the configured default integration branch (choose exactly one of `test` or `main`) | build **and push** | package **and push** |"
            },
            {
              "line": 103,
              "text": "| tag `v*` | no publication by default; use a separately authorized trusted release workflow | no publication by default; use a separately authorized trusted release workflow |"
            },
            {
              "line": 105,
              "text": "A PR must never publish or receive registry credentials. When push and pull-request triggers could validate the same candidate, split the policy into manual tokenless branch readiness, automatic PR validation with one required aggregate result, and trusted default-branch publication. Do not query private PR state from candidate-controlled code with a repository token."
            },
            {
              "line": 107,
              "text": "Publication retries must fail closed: reuse an existing image tag only after its remote OCI config digest matches the validated local image ID, and reuse an existing chart version only after the pulled archive bytes match the deterministic local package. The final aggregate runs under `always()` and explicitly requires every publication job to succeed."
            },
            {
              "line": 109,
              "text": "## 4. Installing the pipeline into a monorepo"
            },
            {
              "line": 112,
              "text": "2. Implement separate credential-free validation and trusted publication workflows when candidate and publication trust boundaries differ."
            },
            {
              "line": 113,
              "text": "3. Keep Harbor credentials only in trusted default-branch publication jobs or a separately authorized isolated release workflow; never at workflow-global scope."
            },
            {
              "line": 114,
              "text": "4. Add contract tests for triggers, exact checkout identity, credential isolation, immutable naming/retry, evidence shape, and forbidden deployment commands."
            },
            {
              "line": 115,
              "text": "5. Verify a branch/PR exact head, then merge and confirm immutable images and charts through direct Harbor readback."
            },
            {
              "line": 117,
              "text": "## Handoff"
            },
            {
              "line": 121,
              "text": "## Gotchas"
            },
            {
              "line": 126,
              "text": "- **Charts are centralized, not co-located.** They live in `helm-charts/<chart>/`. An app can"
            },
            {
              "line": 128,
              "text": "  independently, do not assume a 1:1 pairing."
            },
            {
              "line": 129,
              "text": "- **Gitea has no global secrets \u2014 only global variables.** The username/address are variables"
            },
            {
              "line": 131,
              "text": "  \"global secret\"; the admin Actions page only offers global *variables*."
            },
            {
              "line": 132,
              "text": "- **The robot secret is only shown once, at creation.** Harbor never reveals it again on GET."
            },
            {
              "line": 135,
              "text": "- **`runs-on` maps to a container image \u2014 it must have required tools.** A job that builds images needs Docker and Git; chart publication also needs Helm. A bare `node:alpine` label has no Docker CLI and fails with `docker: not found`."
            },
            {
              "line": 136,
              "text": "- **Marketplace actions may not resolve.** Gitea may proxy `actions/*` to github.com. If the runner has no egress, mirror the pinned action into Gitea or use a repository-approved exact-revision checkout that does not expose a reusable token to candidate code."
            },
            {
              "line": 140,
              "text": "- **`helm push` needs an OCI-enabled Helm \u2265 3.8 and a prior `helm registry login`.** Without"
            },
            {
              "line": 141,
              "text": "  the login step, `helm push` fails with `unauthorized` even though `docker login` succeeded \u2014"
            }
          ]
        },
        {
          "path": "references/pipeline-workflow.md",
          "installed_sha256": "e9a17855ff5e424ceabad2ffde293218fe0ab679fa5719021c898c597667538d",
          "lines": 74,
          "remote_comparison": "not performed for this supplemental file",
          "excerpts": [
            {
              "line": 1,
              "text": "# Secure Gitea Actions pipeline design"
            },
            {
              "line": 3,
              "text": "Use this reference when choosing triggers, separating candidate validation from trusted publication, or implementing immutable Harbor retries and evidence."
            },
            {
              "line": 5,
              "text": "## Trust boundaries"
            },
            {
              "line": 7,
              "text": "Treat candidate-controlled validation and trusted publication as different workflows whenever credentials or repository authority are involved."
            },
            {
              "line": 9,
              "text": "### 1. Branch readiness"
            },
            {
              "line": 11,
              "text": "Use manual `workflow_dispatch` for a non-default branch when automatic private-PR classification would require a repository token in candidate-controlled code. Check out the exact event revision with persisted credentials disabled. Run service-free validation and build all required applications."
            },
            {
              "line": 13,
              "text": "### 2. Pull-request validation"
            },
            {
              "line": 15,
              "text": "Run automatically for PRs targeting the default integration branch. Validate the exact PR head without registry credentials, deployment credentials, databases, running applications, journeys, or deployment commands. Expose one required aggregate result and cancel superseded runs."
            },
            {
              "line": 19,
              "text": "### 3. Integrated publication"
            },
            {
              "line": 21,
              "text": "Publish only from the exact default-branch integration commit. Complete candidate execution before introducing Harbor credentials. Build, validate, publish, read back, and record immutable images and OCI charts. Deployment and GitOps mutation remain outside this workflow."
            },
            {
              "line": 23,
              "text": "## Exact checkout"
            },
            {
              "line": 27,
              "text": "## Immutable identity and retry"
            },
            {
              "line": 31,
              "text": "A retry may reuse an existing image tag only when the remote manifest's OCI config digest equals the validated local image ID. A tag containing different content is an immutability violation and must fail closed."
            },
            {
              "line": 33,
              "text": "Make chart identity source-specific. Package deterministically. A retry may reuse an existing chart version only when the pulled archive bytes hash exactly like the local package; mismatched bytes must fail closed."
            },
            {
              "line": 35,
              "text": "## Evidence model"
            },
            {
              "line": 46,
              "text": "- canonical millisecond ISO-8601 UTC publication timestamp."
            },
            {
              "line": 48,
              "text": "Generate the aggregate result under `always()` and explicitly fail unless every required publication job succeeded. Preserve the validated source URL and timestamp in the final record rather than dropping them after validation."
            },
            {
              "line": 50,
              "text": "## Credential lifecycle"
            },
            {
              "line": 52,
              "text": "Keep registry secrets at publication-job scope, never workflow-global scope. Docker and Helm use separate credential stores. Authenticate through standard input, clean temporary material on success or failure, and treat logout failure as a job failure on persistent runners after cleanup completes."
            },
            {
              "line": 54,
              "text": "## Build context and discovery"
            },
            {
              "line": 56,
              "text": "For a pnpm/Turborepo monorepo, Docker builds generally require the repository root as context so the lockfile and workspace dependencies are available. Discover images and charts independently; never assume a 1:1 mapping unless the repository contract explicitly requires it. Fail if a required discovery set is unexpectedly empty."
            },
            {
              "line": 58,
              "text": "## Verification"
            },
            {
              "line": 60,
              "text": "1. Contract-test trigger separation, exact checkout, credential isolation, all-app coverage, immutable identity/retry, evidence shape, and forbidden deployment commands."
            },
            {
              "line": 61,
              "text": "2. Run local format, lint, typecheck, unit/security, container, chart, and build gates."
            },
            {
              "line": 62,
              "text": "3. Verify one complete exact-head PR attempt and its full job inventory."
            },
            {
              "line": 63,
              "text": "4. After merge, resolve the generated integration commit and verify its own publication jobs."
            },
            {
              "line": 66,
              "text": "## Pitfalls"
            },
            {
              "line": 74,
              "text": "- Warning-only credential cleanup on persistent runners."
            }
          ]
        }
      ],
      "remote_branch": "test",
      "remote_commit": "be07b8183518b20e8302bca06dafe436285803cd",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/software-development/devsecops-ci-cd-gitea/SKILL.md"
    },
    {
      "skill": "devsecops-ci-cd-gitea-aerosim",
      "repository": "corp-v1-aerosim-skills-code-agent/devsecops-ci-cd-gitea-aerosim",
      "installed": true,
      "files": [
        {
          "path": "SKILL.md",
          "installed_sha256": "e8c69a99925102c233ed58297f9fafe73d2aa4b3e6d1f9641641e7cf6cb7a844",
          "lines": 73,
          "remote_sha256": "d2ce0b87c8a20699e92426187081fcb4dfdc539f0d68fefa178fe41b8c85f719",
          "matches_remote": false,
          "remote_url": "https://gitea.lego-cloud.eu/corp-v1-aerosim-skills-code-agent/devsecops-ci-cd-gitea-aerosim/src/commit/c848d57d1d4501a5c597ae559df503d89e97ee84/SKILL.md",
          "excerpts": [
            {
              "line": 5,
              "text": "author: RootAtSkic, Hermes Agent"
            },
            {
              "line": 9,
              "text": "    tags: [aerosim, gitea-actions, ci, harbor, helm]"
            },
            {
              "line": 13,
              "text": "# AeroSim Gitea Actions"
            },
            {
              "line": 16,
              "text": "## Standing AeroSim PR merge authority"
            },
            {
              "line": 18,
              "text": "RootAtSkic authorizes Hermes to merge PRs across all AeroSim repositories and channels without per-PR human authorization. Source: [explicit standing delegation](https://discord.com/channels/1518726359512387766/1537544173366943886/1554873863165182015)."
            },
            {
              "line": 20,
              "text": "This project-local rule supersedes older instructions in this skill, linked references, or inherited guidance requiring a human to authorize PR merging or forbidding Hermes from merging. Preserve it during skill adoption and synchronization. Complete required review and exact-head CI, freshly verify the unchanged mergeable head, merge it, and verify the integration revision and required post-merge checks. Do not block a reviewed, green PR for owner merge approval or request repeated permission."
            },
            {
              "line": 22,
              "text": "This delegates PR integration, not product/design/architecture decisions, human UAT acceptance, or independently gated release/deployment actions. If merging a GitOps or release PR itself triggers a separately gated deployment, establish that deployment authority first; do not relabel it as a PR-merge approval. Repository protections, security checks, and genuine technical blockers still apply."
            },
            {
              "line": 26,
              "text": "Maintain the verified application CI and immutable Harbor publication contract for `corp-v1-aerosim/corp-v1-aerosim`. This project overlay owns AeroSim-specific workflow names, triggers, artifact paths, credential boundaries, and evidence. Deployment remains GitOps-owned."
            },
            {
              "line": 28,
              "text": "## When to use"
            },
            {
              "line": 35,
              "text": "- Web/API image or `charts/aerosim` publication;"
            },
            {
              "line": 38,
              "text": "Read `references/pipeline-workflow.md` before changing triggers, jobs, or publication logic. Read `references/conventions.md` before changing image/chart identities. Read `references/harbor-auth.md` before touching credentials or registry cleanup."
            },
            {
              "line": 40,
              "text": "## Required workflow types"
            },
            {
              "line": 42,
              "text": "| Type | Exact Action name | Trigger | Credentials |"
            },
            {
              "line": 46,
              "text": "| Integrated publication | `CD / Test Artifact Publication` | push to `test` | Harbor credentials only in publication jobs |"
            },
            {
              "line": 48,
              "text": "Branch validation builds every application. Pull-request validation builds every application, builds both container images locally, and lints/packages the Helm chart locally to prove buildability and packageability. Branch and PR validation never authenticate to Harbor, push an image, push an OCI chart, start PostgreSQL or applications, run journeys/GPU/UAT, deploy, or mutate GitOps. Only the trusted integrated `test` workflow may publish artifacts."
            },
            {
              "line": 50,
              "text": "## Publication contract"
            },
            {
              "line": 52,
              "text": "The trusted `test` workflow builds all applications, publishes immutable Web/API images and one source-specific OCI Helm chart, then emits combined provenance evidence. Use the full 40-character source revision for image tags and chart identity."
            },
            {
              "line": 54,
              "text": "A retry may reuse an existing image tag only when its remote OCI config digest equals the validated local image ID. A retry may reuse an existing chart version only when the pulled archive bytes hash exactly like the deterministic local archive. Any mismatch fails closed."
            },
            {
              "line": 56,
              "text": "The final result must run under `always()` and fail unless image and chart publication both succeeded. Preserve source revision, canonical source URL, canonical millisecond UTC timestamp, image references/digests, chart reference/version/digest, and chart archive checksum as distinct fields."
            },
            {
              "line": 58,
              "text": "## Verification"
            },
            {
              "line": 60,
              "text": "1. Run the application repository's CI, container, and Helm contract suites."
            },
            {
              "line": 61,
              "text": "2. Verify exact-head PR validation has one required aggregate result."
            },
            {
              "line": 62,
              "text": "3. After merge, resolve the generated `test` commit and require all publication jobs to pass for that exact revision."
            },
            {
              "line": 66,
              "text": "## Pitfalls"
            },
            {
              "line": 70,
              "text": "- Short SHA or mutable `latest` publication."
            },
            {
              "line": 72,
              "text": "- Rejecting a byte-identical partial publication retry or accepting a mismatched immutable artifact."
            },
            {
              "line": 73,
              "text": "- Warning-only Docker/Helm logout on persistent runners."
            }
          ]
        },
        {
          "path": "references/pipeline-workflow.md",
          "installed_sha256": "15379afb16c6d301d517e51eb113c4fda0b632b9548d3a8fe2f8dbe5cda63b53",
          "lines": 66,
          "remote_comparison": "not performed for this supplemental file",
          "excerpts": [
            {
              "line": 1,
              "text": "# AeroSim three-case pipeline"
            },
            {
              "line": 3,
              "text": "## 1. Non-default branch without a pull request"
            },
            {
              "line": 5,
              "text": "`CI / Branch Readiness` is manually dispatched on the exact branch commit. It checks out `${{ gitea.sha }}` with persisted credentials disabled, installs once, runs formatting, lint, type checking and service-free unit tests, then builds every application."
            },
            {
              "line": 7,
              "text": "Manual dispatch is intentional: automatically classifying whether a private branch has a PR would require API authority in candidate-controlled workflow code. Do not expose a repository token for that classification. Use branch-scoped concurrency with cancellation so a newer manual run cancels the superseded Branch Readiness run for the same branch."
            },
            {
              "line": 9,
              "text": "## 2. Pull request targeting `test`"
            },
            {
              "line": 11,
              "text": "`CI / Pull Request Validation` runs automatically for the exact PR head. It performs service-free unit validation and an all-application build, then builds the Web and API container images locally and lints/packages the Helm chart locally to prove the exact candidate is buildable and packageable. The candidate workflow must use no Harbor or registry credentials and must never log in, push, publish, deploy, or mutate GitOps. Its locally built images and packaged chart are disposable validation outputs, not release artifacts. One `always()` aggregate job is the required merge result and explicitly requires both code validation and local image/chart packaging to succeed."
            },
            {
              "line": 15,
              "text": "## 3. Push to `test`"
            },
            {
              "line": 17,
              "text": "`CD / Test Artifact Publication` runs only for the exact integrated commit:"
            },
            {
              "line": 20,
              "text": "build all applications"
            },
            {
              "line": 23,
              "text": "        \u2514\u2500\u2500 package, publish, and read back the OCI Helm chart"
            },
            {
              "line": 24,
              "text": "            \u2514\u2500\u2500 validate combined publication evidence"
            },
            {
              "line": 27,
              "text": "Candidate execution is complete before Harbor credentials are introduced. The workflow does not run services, journeys, GPU qualification, deployment commands, or GitOps mutations."
            },
            {
              "line": 29,
              "text": "The validated image archive crosses job boundaries only through bounded, deterministically named chunks plus a checksum and source-bound eligibility record. Reconstruct it in a visible workspace directory, reject extra/missing/symlinked/out-of-order chunks, verify the checksum, load the archive, and verify both loaded image identities and source provenance before registry credentials are accepted."
            },
            {
              "line": 31,
              "text": "Contract tests must fail if branch/PR jobs reference Harbor credentials, if publication credentials are available before validated artifact reconstruction, or if any workflow introduces `kubectl`, `helm upgrade`, `argocd`, kubeconfig material, direct GitOps pushes, mutable deployment tags, or unresolved dynamic shell execution."
            },
            {
              "line": 33,
              "text": "## Immutable identity and retry"
            },
            {
              "line": 35,
              "text": "- Image tag: exact 40-character source revision."
            },
            {
              "line": 36,
              "text": "- Image retry: inspect the remote manifest's OCI config digest and require equality with the validated local image ID before treating an existing tag as complete."
            },
            {
              "line": 41,
              "text": "## Evidence"
            },
            {
              "line": 53,
              "text": "Combined evidence must retain the canonical repository URL and a canonical millisecond ISO-8601 UTC publication timestamp. The publication-result job uses `always()` and fails unless both image and chart publication jobs succeeded."
            },
            {
              "line": 55,
              "text": "## Credential lifecycle"
            },
            {
              "line": 57,
              "text": "Docker and Helm use separate credential stores. Authenticate only inside trusted publication jobs. Cleanup runs after success or failure; on persistent runners, a logout failure is a job failure after temporary files are removed, not a warning."
            },
            {
              "line": 59,
              "text": "## Validation sequence"
            },
            {
              "line": 61,
              "text": "1. Contract-test the three exact workflow names, triggers, job graph, and forbidden runtime/deployment commands."
            },
            {
              "line": 62,
              "text": "2. Run format, lint, typecheck, unit, security, CI contract, container contract, Helm contract, all-app build, dependency audit, and `git diff --check`."
            },
            {
              "line": 63,
              "text": "3. Review the complete exact tree."
            },
            {
              "line": 64,
              "text": "4. Verify exact-head PR CI."
            },
            {
              "line": 65,
              "text": "5. After merge, resolve the new `test` head and verify all publication jobs for that exact revision."
            }
          ]
        }
      ],
      "remote_branch": "test",
      "remote_commit": "c848d57d1d4501a5c597ae559df503d89e97ee84",
      "loadable_installed": true,
      "installed_source_path": "/opt/data/skills/software-development/devsecops-ci-cd-gitea-aerosim/SKILL.md"
    },
    {
      "skill": "development-gitops-argo-cd-gondor-v1",
      "repository": "home-v1-skills-code-agent/development-gitops-argo-cd-gondor-v1",
      "installed": false,
      "files": [
        {
          "path": "SKILL.md",
          "remote_sha256": "6fbcde5994932910b1c46f2956f8e09c4c90fb42895a0e00d282821ac094aabd",
          "remote_url": "https://gitea.lego-cloud.eu/home-v1-skills-code-agent/development-gitops-argo-cd-gondor-v1/src/commit/3c3d495417993d5c15bb3379bdae5c85f04e1890/SKILL.md",
          "excerpts": [
            {
              "line": 7,
              "text": "  cluster/0300-applications and its cluster/0500-resources sibling, wiring a Helm chart via the"
            },
            {
              "line": 27,
              "text": "# Argo CD GitOps \u2014 gondor-v1 instance"
            },
            {
              "line": 39,
              "text": "## Repositories"
            },
            {
              "line": 51,
              "text": "Project adoptions may render the same committed template into **one repository per environment**. When an adopted project skill names separate development and production repositories, that explicit mapping overrides this home-v1 singleton example. Each rendered repository must carry its own project-scoped AppProject and app-of-apps, point only to itself, and fail closed against other environment namespaces, hostnames, resource names, and repository URLs. Never infer that two environments belong in one rendered repository from the `gondor-v1` example."
            },
            {
              "line": 53,
              "text": "## What is different from the base skill \u2014 read before editing"
            },
            {
              "line": 63,
              "text": "| Everything under `cluster/` is `.jinja` | **ESO manifests are plain `.yml` on purpose** so ESO's own `{{ }}` survives (see below) |"
            },
            {
              "line": 68,
              "text": "## The three tiers"
            },
            {
              "line": 77,
              "text": "\u2514\u2500\u2500 0500-resources/                    # Helm values OR raw manifests (base skill's cluster/resources/)"
            },
            {
              "line": 84,
              "text": "## Two Application archetypes"
            },
            {
              "line": 88,
              "text": "1. **Multi-source Helm** \u2014 upstream chart + values pulled from the gondor-v1 repo via a"
            },
            {
              "line": 96,
              "text": "## The cross-repo contract with palantir-v1 \u2014 the thing most likely to bite"
            },
            {
              "line": 115,
              "text": "## Scope routing"
            },
            {
              "line": 119,
              "text": "| Generic Argo/Copier/Helm mechanics, the three-layer contract, validation commands | base skill `development-gitops-argo-cd` + its references |"
            },
            {
              "line": 123,
              "text": "## Gotchas specific to gondor-v1"
            },
            {
              "line": 129,
              "text": "  verbatim and ESO's own `{{ \u2026 }}` template survives. Renaming one to `.yml.jinja` makes Copier"
            },
            {
              "line": 132,
              "text": "- **The Helm `valueFiles` path is hardcoded to the resources dir.** A multi-source app's"
            },
            {
              "line": 134,
              "text": "  exactly. Rename the 0300 dir without renaming the 0500 dir (or vice-versa) and the app can't"
            },
            {
              "line": 136,
              "text": "- **Storage/secrets are a prerequisite, not part of the sync.** Before enabling `harbor`,"
            },
            {
              "line": 139,
              "text": "- **Argo CD is exposed by its own NodePort, not the ingress.** Every other app is reached at"
            }
          ]
        }
      ],
      "remote_branch": "test",
      "remote_commit": "3c3d495417993d5c15bb3379bdae5c85f04e1890",
      "loadable_installed": false,
      "inspection_basis": "Pinned remote SKILL.md; skill_view lookup unavailable. No substitution or installation performed."
    }
  ],
  "coverage": {
    "internal_references": 9,
    "declared_external_entry_points": 18,
    "note": "All nine internal references read. Findings compare relevant sections of 18 declared external skill entry points and selected supporting references, not every transitive dependency."
  }
}
