[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M72XV405VR6Y5AFVDTYA0QSK":3,"skill-same:01M72XV405VR6Y5AFVDTYA0QSK":79},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":27,"commitSha":28,"skillDirectory":29,"forksCount":30,"repositoryHtmlUrl":31,"repositoryHomepageUrl":32,"ownerHtmlUrl":33,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":34,"install":35,"resources":48},"01M72XV405VR6Y5AFVDTYA0QSK","hex-plan",null,"Tiered multi-agent planning orchestrator — decomposes a feature, issue, or PR into a contract-first TDD plan through discover, research, design, decompose, and adversarial-review phases. Use for multi-step feature planning, task decomposition, swarm planning, multi-perspective research, meta-plan preview, or ADR scaffolding. Tier (low|medium|high|xhigh|max, auto by default) scales research depth, whether an architect designs, and review breadth.","grimoire-rs\u002Fgrimoire","grimoire-rs","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F298895348?v=4","01M72XK7XKYKMET2MMHCAQ3X6C","01M72XK7Y647QCPS3Z02ZYTJTT",15,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhex-plan\u002FSKILL.md",[],[],"2026-11-10 22:39:42",false,"---\nname: hex-plan\ndescription: Tiered multi-agent planning orchestrator — decomposes a feature, issue, or PR into a contract-first TDD plan through discover, research, design, decompose, and adversarial-review phases. Use for multi-step feature planning, task decomposition, swarm planning, multi-perspective research, meta-plan preview, or ADR scaffolding. Tier (low|medium|high|xhigh|max, auto by default) scales research depth, whether an architect designs, and review breadth.\nlicense: Apache-2.0\nmetadata:\n  summary: Tiered swarm planning — prompt\u002Fissue to reviewed contract-first plan\n  keywords: planning,swarm,multi-agent,decomposition,research,adr,tdd\n  repository: https:\u002F\u002Fgithub.com\u002Fmichael-herwig\u002Farcana\n---\n\n# hex-plan — Planning Orchestrator\n\nThin dispatcher. It parses arguments, classifies the target's tier, resolves\noverlays, runs the single meta-plan approval gate, announces the resolved\nconfig, and hands off to the matching tier file. The phase plans live in\nthe five `tier-\u003Ctier>.md` files; the shared vocabulary\n(tiers, the Review-Fix Loop, worker roles, model classes, the memory file)\nlives in the `hex-core` reference library and is **linked here, never\ncopied**.\n\nShared contracts:\n[`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md) ·\n[`workers.md`](..\u002Fhex-core\u002Freferences\u002Fworkers.md) ·\n[`models.md`](..\u002Fhex-core\u002Freferences\u002Fmodels.md) ·\n[`memory.md`](..\u002Fhex-core\u002Freferences\u002Fmemory.md).\nIf `hex-core` is not installed: `grim add ghcr.io\u002Fmichael-herwig\u002Farcana\u002Fhex-core:latest`.\n\n## Argument syntax\n\n```\n\u002Fhex-plan [tier] \u003Ctarget> [flags]\n```\n\n- **tier** (optional): `low | medium | high | xhigh | max | auto`. Default\n  `auto` — the classifier picks the lowest-fit tier, `low` … `xhigh`\n  ([`classify.md`](classify.md)); a higher tier needs the user. **`max` is explicit\n  only** — `--tier=max`, or a plan whose Status block says `Tier: max`; the\n  classifier never emits it\n  ([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#tier-grammar)).\n- **target** (one of):\n  - free-text prompt;\n  - `\u003CN>` or `#\u003CN>` — a number; probe PR first, fall back to issue;\n  - `PR \u003CN>` \u002F `pull\u002F\u003CN>` — an explicit pull request;\n  - `issue \u003CN>` \u002F `issues\u002F\u003CN>` — an explicit issue;\n  - a full GitHub URL `https:\u002F\u002Fgithub.com\u002F\u003Cowner>\u002F\u003Crepo>\u002F(pull|issues)\u002F\u003CN>`.\n- **flags** (before the target, by convention):\n  - `--architect=inline|on` — override the Design phase (see\n    [`overlays.md`](overlays.md)).\n  - `--research=skip|1|3` — override Research breadth.\n  - `--adversary` \u002F `--no-adversary` — force the cross-model plan-artifact\n    pass on or off.\n  - `--dry-run` — make the meta-plan gate block for explicit approval and\n    stop there (preview only, no workers launched).\n\n## Dispatch\n\nThe outer loop every hex orchestrator shares\n([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#shared-shape)):\n\n### 1. Parse arguments and resolve memory\n\nParse the tier, target, and flags. Locate `.agents\u002Fmemory\u002Fhex.md` by\nsearching upward from the working directory; a missing file is normal\n([`memory.md`](..\u002Fhex-core\u002Freferences\u002Fmemory.md#location-and-resolution)) —\nfall back to shipped defaults and note that `\u002Fhex-init` can create one. When\npresent, read `hex.md › Pointers` for cached locations (where verification\nand plan\u002FADR conventions are documented; verify on consumption, re-point on\na miss), `hex.md › Preferences` for the instantiated model matrix, the\nadversary skill name, limits, and always-on perspectives, and\n`hex.md › Memory` for any active plan already in flight. Product context\n(what the product is, users, related repos, research keywords) lives in\nproject context, located via `hex.md › Pointers`.\n\nIf the resolved `hex.md` carries a `Federation lead:` bullet, **halt** per\n[`memory.md` § Location and resolution](..\u002Fhex-core\u002Freferences\u002Fmemory.md#location-and-resolution)\n— this repo is a federation satellite and its memory is not the plan's.\n\n### 2. Resolve the target\n\nIf the target is a GitHub reference, fetch its title, body, labels, and\n(for PRs) file list, in this order of preference:\n\n1. the client's **GitHub MCP tools** (PR\u002Fissue read) when it exposes them;\n2. the **`gh` CLI** (`gh pr view` \u002F `gh issue view --json\n   title,body,comments,labels,files`) as fallback;\n3. neither available — treat the input as **free text** and continue.\n\nProbe order for a bare `\u003CN>`\u002F`#\u003CN>` is a PR then an issue; PRs and issues are\nequal-class targets. The fetched body and labels feed classification; a PR's\nfile list feeds the Discover scope, not the tier decision.\n\n### 3. Classify (only when tier is `auto`)\n\nRead [`classify.md`](classify.md). Apply its tier signals and overlay\ntriggers to the prompt plus any fetched GitHub context. It emits a candidate\ntier (**only** `low`, `medium`, `high`, or `xhigh`), a confidence flag, and an overlay\nset. Low confidence forces the gate in step 5. Never ask a mid-flow question\nduring classification — ambiguity is resolved at the single gate.\n\n### 4. Resolve overlays\n\nFinal config = the tier's baseline from [`overlays.md`](overlays.md) +\nclassifier-inferred overlays + `hex.md › Preferences` hints + user flags.\nLater wins;\n**user flags always override** (see\n[`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#spawn-selection-precedence)).\n\n### 5. Meta-plan gate (the single approval point)\n\nExactly one gate, before any worker launches\n([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#the-meta-plan-approval-gate)).\nIts weight scales:\n\n- **Confident `low` \u002F `medium` \u002F `high`** — announce the resolved config (step 6) and\n  proceed; the user can still abort. An explicit user tier always counts as\n  confident — the classifier never ran.\n- **`xhigh` or `max` tier, low-confidence classification, or `--dry-run`** — block for\n  explicit approval.\n\nOn a client with a **native plan-approval mechanism**, use it. Otherwise\npresent the announce block as **one structured question** (approve \u002F adjust \u002F\ncancel). The gate is also where the user can add or drop perspectives, choose\nresearch axes, and adjust models. On adjust, re-resolve and re-present once.\nNever split this into sequential questions. Open questions are presented\ntogether with their `Recommended: \u003Canswer> — \u003Creason>` line; a plain\napproval accepts every recommendation as-is\n([the meta-plan approval gate](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#the-meta-plan-approval-gate)).\n\n### 6. Announce the resolved config\n\nPrint the resolved config with per-axis source attribution before loading the\ntier file (format:\n[`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#the-meta-plan-approval-gate)):\n\n```\nhex-plan\n  Tier:      high                         (auto — classifier: new subcommand, 2 areas)\n  Overlays:  architect=inline               (accepted ADR covers the design)\n             research=1                      (tier baseline)\n             adversary=off                   (tier baseline)\n  Spawn set:\n    architecture-explorer                    (tier baseline)\n    explorer ×3                              (tier baseline)\n    researcher ×1                            (overlay research=1)\n    reviewer: spec                           (tier baseline)\n  Models:    standard default                                    (models.md)\n  Adversary: off\n  Degraded:  no — subagent spawning available\n```\n\nThe announce block prints one config-disclosure line per change a\n`hex.md › Preferences` block makes; the full trigger set is defined once in\n[`protocol.md` § the meta-plan approval gate](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#the-meta-plan-approval-gate).\nFor this skill, for example:\n\n```\n  [project-redefined: Research.researcher 1→3 (hex.md tiers)]\n  [architect dropped — phase ceiling 3 reached]\n  [Review batched 2+2 — concurrency cap 4 (hex.md)]\n  Error: never [reviewer:security] refused (hex.md preference)\n  Fix:   see config.md#merge-rules for the attestation requirement\n```\n\nEvery spawn line carries its source (`tier baseline` \u002F `classifier` \u002F\n`hex.md preference` \u002F `user flag`) per\n[spawn-selection precedence](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#spawn-selection-precedence).\nModel names above are class placeholders — shipped files never hardcode\nliterals; the running orchestrator resolves and prints each spawn's\nliteral model per the\n[Models line contract](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#the-meta-plan-approval-gate). On a client that cannot spawn subagents, announce\n`Degraded: inline workers` and run each worker prompt inline and sequentially\n([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#worker-coordination)).\n\n### 7. Dispatch to the tier file\n\nRead `workflows.hex-plan.\u003Ctier>` from the resolved config first. When set and\nthe named file passes the seven validation checks\n([`config.md` § Workflows](..\u002Fhex-core\u002Freferences\u002Fconfig.md#workflows)), `Read`\nthat forked file in place of the shipped one; on validation failure or when\nunset, `Read` the matching `tier-{low,medium,high,xhigh,max}.md` — config.md owns the\ncheck list and the on-failure fallback. Announce which ran: `Workflow: shipped\ntier-\u003Ctier>.md`, or for a fork, `Workflow: \u003Cpath> (forked from \u003Cshipped tier\nfile> @ \u003Cstamped version>)`. Execute the loaded file's phase plan; no phase\ncontent is duplicated in this file.\n\n## Worker assignment (shared across tiers)\n\nRoles are indexed in [`workers.md`](..\u002Fhex-core\u002Freferences\u002Fworkers.md);\nthe orchestrator loads a full persona (spawn-prompt template, output\ncontract) only for roles in the resolved spawn set\n([spawn-selection precedence](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#spawn-selection-precedence)).\nThe model class for each role × tier is in\n[`models.md`](..\u002Fhex-core\u002Freferences\u002Fmodels.md). This table maps roles to\nplanning phases; the tier files set the actual counts.\n\n| Phase | Role | Count | Purpose |\n|---|---|---|---|\n| Discover | `architecture-explorer` | 0–1 | Map the current architecture, dependencies, reusable code |\n| Discover | `explorer` | 1–4 | Deep-dive each involved area |\n| Research | `researcher` | 0–1 (3 only when the user asks) | Technology \u002F patterns \u002F domain landscape |\n| Design | `architect` | 0–1 | ADR or system design — only for a new one-way-door decision no accepted ADR covers |\n| Review | `reviewer` (focus `spec`) | 1 | Plan ↔ design consistency; for a plan built from an accepted ADR, the decomposition only |\n| Review | `architect` | 0–1 | Trade-off honesty (new one-way door with no ADR, or the user asks) |\n| Review | `researcher` | 0–1 | SOTA \u002F known-pitfall gap check (same condition) |\n| Adversary | configured adversary skill (`plan-artifact`) | 0–1 | Cross-model review (same condition) |\n\nA project's `tiers.hex-plan.\u003Ctier>.counts` can override any Count cell above\nagainst the baseline this table sets\n([`config.md` § Merge rules](..\u002Fhex-core\u002Freferences\u002Fconfig.md#merge-rules)).\n\nConcurrency cap and degraded mode:\n[`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#worker-coordination). The\nadversary skill name comes from `hex.md › Preferences`\n([adversary contract](..\u002Fhex-core\u002Freferences\u002Fadversary.md#adversary-contract));\n`codex-adversary` is only an example value.\n\n## Project rules and conventions\n\nhex never carries a hardcoded rule or subsystem table. Every phase that needs\nthe project's invariants, verification command, or artifact conventions\ndiscovers them from **project context** (the client's ambient instructions \u002F\nproject rules), cached in the Pointers section of\n`.agents\u002Fmemory\u002Fhex.md`\n([`memory.md`](..\u002Fhex-core\u002Freferences\u002Fmemory.md#the-three-sections)). \"Verify\"\nanywhere below means **run the project's documented verification**\n([`verify.md`](..\u002Fhex-core\u002Freferences\u002Fverify.md#verification)).\n\n## The plan artifact\n\n**Location.** Write the plan where the project's documented plan\u002FADR\nconventions say (discovered from project context, cached in\n`hex.md › Pointers`). When nothing is documented, use the default artifact\nhome `.agents\u002Fplans\u002F`\n([`memory.md`](..\u002Fhex-core\u002Freferences\u002Fmemory.md#location-and-resolution)).\nRecord the active-plan pointer in `hex.md › Memory`. The plan template shipped\nwith `\u002Fhex-init` is the scaffold; the project's own format wins when it has\none. A plan carrying a `Repo` column lives in the **lead only** — one\nartifact, one Status block — with its active-plan pointer recorded solely in\nthe lead's `hex.md › Memory`\n([`memory.md`](..\u002Fhex-core\u002Freferences\u002Fmemory.md#federation-satellites)).\n\n**Status block.** Every plan opens with a self-contained status block the\nexecute and review skills read and mutate — no external state file:\n\n```markdown\n## Status\n- State:   plan-approved      \u003C!-- planning → plan-approved → executing → review → done -->\n- Tier:    high\n- Tier-grammar: 5\n- Updated: 2026-07-19\n- Next:    \u002Fhex-execute \u003Cthis plan path>\n```\n\nhex-plan initializes it at `plan-approved` on handoff, writes\n`- Tier-grammar: 5` (the tier grammar the `Tier:` value was written in,\n[`protocol.md` § Tier grammar](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#tier-grammar),\n`adr_0017` C-997) into every new plan's Status block, and records the pointer\nin `hex.md › Memory`; `\u002Fhex-execute` advances `State` and `Next` as it runs.\n\n**Required content** (every tier — the tier files scale depth, not presence):\n\n- **Component contracts** — the public surface touched (types, signatures,\n  error variants) with expected behavior and edge cases, testable enough that\n  a tester could write failing tests from them without reading any code.\n  Numbered `C-001, C-002, …`; UX scenarios below are numbered\n  `S-001, S-002, …` — both are the coverage join keys checked at review and\n  convergence\n  ([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#traceability-ids)).\n- **User-experience scenarios** — action → expected outcome → error cases for\n  each user-facing behavior.\n- **Contract wave** — the stubs **plus contract tests** for every pipeline,\n  committed once before any pipeline starts. Contract-first TDD runs inside\n  each step (Stub → Specify → Implement), never as plan-level phases. A\n  one-pipeline plan has no contract wave.\n- **Pipelines** — the plan is a few **pipelines** cut along contracts, each an\n  ordered chain of **steps**; cutting rules and the table's columns are in\n  [`decompose.md`](..\u002Fhex-core\u002Freferences\u002Fdecompose.md#parallel-by-default-decomposition).\n  A Parallelization table with one row per pipeline (id, repo, scope citing\n  the C-\u002FS- IDs it covers, expected files, wave, depends-on, marks, status —\n  initialized `pending`), the steps of each pipeline as an ordered list of\n  small briefs, a wave-grouped mermaid `graph TD` as its visual index (the\n  table stays canonical), the critical path, a \"Shippable after wave: N\"\n  line (tier medium and below exempt — one pipeline), and — when fewer\n  parallel pipelines than file-disjointness allows — a one-line\n  justification\n  ([`worktree.md`](..\u002Fhex-core\u002Freferences\u002Fworktree.md#pipeline-worktree-mechanics)).\n- **Marks** — rare, set only where the author knows better than the defaults;\n  `hard` on at most one pipeline in four, `review` on a pipeline that earns\n  its own review\n  ([`decompose.md`](..\u002Fhex-core\u002Freferences\u002Fdecompose.md#parallel-by-default-decomposition),\n  [`models.md`](..\u002Fhex-core\u002Freferences\u002Fmodels.md)). Everything else runs\n  `standard`.\n- **Open questions** — unresolved ambiguities as `[NEEDS CLARIFICATION: …]`\n  markers, **hard cap 3**. More than three means the target is underspecified;\n  raise it at the gate rather than guessing.\n\n## Constraints\n\n- Every task carries **testable acceptance criteria** — no vague behaviors.\n- **Discover runs at every tier** — never assume context.\n- **Each artifact is reviewed once; research is never repeated.** A plan\n  built from an accepted ADR (or a discussion handed off with one) never\n  re-runs research or an architect on that ADR's decisions, and its single\n  reviewer checks only the decomposition — contracts testable, pipelines cut\n  cleanly along contracts, steps sized for a fresh agent\n  ([plan-artifact scope](..\u002Fhex-core\u002Freferences\u002Floop.md#the-review-fix-loop)).\n- **Never skip Review**; never exceed the concurrency cap\n  ([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#worker-coordination)).\n- **No mid-flow questions** — ambiguity is resolved at the single gate.\n- **Persist substantial research** (more than a paragraph) as a research\n  artifact in the convention-resolved location.\n- **Always** include component contracts, UX scenarios with error cases, and\n  the Parallelization section.\n- **Always** initialize the plan's Status block and record the active-plan\n  pointer in `hex.md › Memory`.\n- **Constitution gate** — when project context names a constitution (cached\n  in `hex.md › Pointers`), the Design and Review phases gate the plan\n  against it\n  ([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#constitution-gate));\n  violations need a Constitution Deviations row.\n- **Never commit and never push** — this skill plans only.\n- **Upkeep** ([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#upkeep-step)):\n  as the final phase, re-point any `hex.md › Pointers` entry this run\n  revealed as drifted (a changed verification command, a new artifact home)\n  and update `hex.md › Memory`; a perspective gap belongs in\n  `hex.md › Memory` as a note for the next `\u002Fhex-init` run, never written to\n  `hex.md › Preferences` directly.\n\n## Handoff\n\nThe [handoff contract](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#handoff-contract)\napplies: this block is the run's required final message; one optional\nproceed question may follow it.\n\n```markdown\n## Plan Complete: \u003Cfeature | \"Resolves #N\">\n\n### Classification\n- Scope: small | medium | large\n- Reversibility: two-way | one-way (medium) | one-way (high)\n- Tier: low | medium | high | xhigh | max\n- Overlays: architect=\u003Cinline|on>, research=\u003Cskip|1|3>, adversary=\u003Con|off>\n\n### Artifacts\n- \u003Cplan path> (Status block initialized; active-plan pointer in hex.md › Memory)\n- \u003Cresearch artifact path(s)>\n- \u003CADR path> (one-way-door only)\n\n### Pipelines (for \u002Fhex-execute)\n- Contract wave: stubs and contract tests to commit first (none for one pipeline)\n- Parallelization: \u003CN> pipelines, \u003CM> steps in \u003CW> waves; critical path \u003Cpipeline a → pipeline b>; marks: \u003CK> `hard`, \u003CJ> `review`\n\n### Deferred findings (need human judgment)\n- Review panel: …\n- Cross-model review: …\n\n### Next step\n    \u002Fhex-execute \u003Cplan path>\n```\n\nConsumers: `\u002Fhex-execute` (the plan artifact); the human (deferred findings).\n\n$ARGUMENTS\n","# hex-plan — Planning Orchestrator\n\nThin dispatcher. It parses arguments, classifies the target's tier, resolves\noverlays, runs the single meta-plan approval gate, announces the resolved\nconfig, and hands off to the matching tier file. The phase plans live in\nthe five `tier-\u003Ctier>.md` files; the shared vocabulary\n(tiers, the Review-Fix Loop, worker roles, model classes, the memory file)\nlives in the `hex-core` reference library and is **linked here, never\ncopied**.\n\nShared contracts:\n[`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md) ·\n[`workers.md`](..\u002Fhex-core\u002Freferences\u002Fworkers.md) ·\n[`models.md`](..\u002Fhex-core\u002Freferences\u002Fmodels.md) ·\n[`memory.md`](..\u002Fhex-core\u002Freferences\u002Fmemory.md).\nIf `hex-core` is not installed: `grim add ghcr.io\u002Fmichael-herwig\u002Farcana\u002Fhex-core:latest`.\n\n## Argument syntax\n\n```\n\u002Fhex-plan [tier] \u003Ctarget> [flags]\n```\n\n- **tier** (optional): `low | medium | high | xhigh | max | auto`. Default\n  `auto` — the classifier picks the lowest-fit tier, `low` … `xhigh`\n  ([`classify.md`](classify.md)); a higher tier needs the user. **`max` is explicit\n  only** — `--tier=max`, or a plan whose Status block says `Tier: max`; the\n  classifier never emits it\n  ([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#tier-grammar)).\n- **target** (one of):\n  - free-text prompt;\n  - `\u003CN>` or `#\u003CN>` — a number; probe PR first, fall back to issue;\n  - `PR \u003CN>` \u002F `pull\u002F\u003CN>` — an explicit pull request;\n  - `issue \u003CN>` \u002F `issues\u002F\u003CN>` — an explicit issue;\n  - a full GitHub URL `https:\u002F\u002Fgithub.com\u002F\u003Cowner>\u002F\u003Crepo>\u002F(pull|issues)\u002F\u003CN>`.\n- **flags** (before the target, by convention):\n  - `--architect=inline|on` — override the Design phase (see\n    [`overlays.md`](overlays.md)).\n  - `--research=skip|1|3` — override Research breadth.\n  - `--adversary` \u002F `--no-adversary` — force the cross-model plan-artifact\n    pass on or off.\n  - `--dry-run` — make the meta-plan gate block for explicit approval and\n    stop there (preview only, no workers launched).\n\n## Dispatch\n\nThe outer loop every hex orchestrator shares\n([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#shared-shape)):\n\n### 1. Parse arguments and resolve memory\n\nParse the tier, target, and flags. Locate `.agents\u002Fmemory\u002Fhex.md` by\nsearching upward from the working directory; a missing file is normal\n([`memory.md`](..\u002Fhex-core\u002Freferences\u002Fmemory.md#location-and-resolution)) —\nfall back to shipped defaults and note that `\u002Fhex-init` can create one. When\npresent, read `hex.md › Pointers` for cached locations (where verification\nand plan\u002FADR conventions are documented; verify on consumption, re-point on\na miss), `hex.md › Preferences` for the instantiated model matrix, the\nadversary skill name, limits, and always-on perspectives, and\n`hex.md › Memory` for any active plan already in flight. Product context\n(what the product is, users, related repos, research keywords) lives in\nproject context, located via `hex.md › Pointers`.\n\nIf the resolved `hex.md` carries a `Federation lead:` bullet, **halt** per\n[`memory.md` § Location and resolution](..\u002Fhex-core\u002Freferences\u002Fmemory.md#location-and-resolution)\n— this repo is a federation satellite and its memory is not the plan's.\n\n### 2. Resolve the target\n\nIf the target is a GitHub reference, fetch its title, body, labels, and\n(for PRs) file list, in this order of preference:\n\n1. the client's **GitHub MCP tools** (PR\u002Fissue read) when it exposes them;\n2. the **`gh` CLI** (`gh pr view` \u002F `gh issue view --json\n   title,body,comments,labels,files`) as fallback;\n3. neither available — treat the input as **free text** and continue.\n\nProbe order for a bare `\u003CN>`\u002F`#\u003CN>` is a PR then an issue; PRs and issues are\nequal-class targets. The fetched body and labels feed classification; a PR's\nfile list feeds the Discover scope, not the tier decision.\n\n### 3. Classify (only when tier is `auto`)\n\nRead [`classify.md`](classify.md). Apply its tier signals and overlay\ntriggers to the prompt plus any fetched GitHub context. It emits a candidate\ntier (**only** `low`, `medium`, `high`, or `xhigh`), a confidence flag, and an overlay\nset. Low confidence forces the gate in step 5. Never ask a mid-flow question\nduring classification — ambiguity is resolved at the single gate.\n\n### 4. Resolve overlays\n\nFinal config = the tier's baseline from [`overlays.md`](overlays.md) +\nclassifier-inferred overlays + `hex.md › Preferences` hints + user flags.\nLater wins;\n**user flags always override** (see\n[`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#spawn-selection-precedence)).\n\n### 5. Meta-plan gate (the single approval point)\n\nExactly one gate, before any worker launches\n([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#the-meta-plan-approval-gate)).\nIts weight scales:\n\n- **Confident `low` \u002F `medium` \u002F `high`** — announce the resolved config (step 6) and\n  proceed; the user can still abort. An explicit user tier always counts as\n  confident — the classifier never ran.\n- **`xhigh` or `max` tier, low-confidence classification, or `--dry-run`** — block for\n  explicit approval.\n\nOn a client with a **native plan-approval mechanism**, use it. Otherwise\npresent the announce block as **one structured question** (approve \u002F adjust \u002F\ncancel). The gate is also where the user can add or drop perspectives, choose\nresearch axes, and adjust models. On adjust, re-resolve and re-present once.\nNever split this into sequential questions. Open questions are presented\ntogether with their `Recommended: \u003Canswer> — \u003Creason>` line; a plain\napproval accepts every recommendation as-is\n([the meta-plan approval gate](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#the-meta-plan-approval-gate)).\n\n### 6. Announce the resolved config\n\nPrint the resolved config with per-axis source attribution before loading the\ntier file (format:\n[`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#the-meta-plan-approval-gate)):\n\n```\nhex-plan\n  Tier:      high                         (auto — classifier: new subcommand, 2 areas)\n  Overlays:  architect=inline               (accepted ADR covers the design)\n             research=1                      (tier baseline)\n             adversary=off                   (tier baseline)\n  Spawn set:\n    architecture-explorer                    (tier baseline)\n    explorer ×3                              (tier baseline)\n    researcher ×1                            (overlay research=1)\n    reviewer: spec                           (tier baseline)\n  Models:    standard default                                    (models.md)\n  Adversary: off\n  Degraded:  no — subagent spawning available\n```\n\nThe announce block prints one config-disclosure line per change a\n`hex.md › Preferences` block makes; the full trigger set is defined once in\n[`protocol.md` § the meta-plan approval gate](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#the-meta-plan-approval-gate).\nFor this skill, for example:\n\n```\n  [project-redefined: Research.researcher 1→3 (hex.md tiers)]\n  [architect dropped — phase ceiling 3 reached]\n  [Review batched 2+2 — concurrency cap 4 (hex.md)]\n  Error: never [reviewer:security] refused (hex.md preference)\n  Fix:   see config.md#merge-rules for the attestation requirement\n```\n\nEvery spawn line carries its source (`tier baseline` \u002F `classifier` \u002F\n`hex.md preference` \u002F `user flag`) per\n[spawn-selection precedence](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#spawn-selection-precedence).\nModel names above are class placeholders — shipped files never hardcode\nliterals; the running orchestrator resolves and prints each spawn's\nliteral model per the\n[Models line contract](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#the-meta-plan-approval-gate). On a client that cannot spawn subagents, announce\n`Degraded: inline workers` and run each worker prompt inline and sequentially\n([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#worker-coordination)).\n\n### 7. Dispatch to the tier file\n\nRead `workflows.hex-plan.\u003Ctier>` from the resolved config first. When set and\nthe named file passes the seven validation checks\n([`config.md` § Workflows](..\u002Fhex-core\u002Freferences\u002Fconfig.md#workflows)), `Read`\nthat forked file in place of the shipped one; on validation failure or when\nunset, `Read` the matching `tier-{low,medium,high,xhigh,max}.md` — config.md owns the\ncheck list and the on-failure fallback. Announce which ran: `Workflow: shipped\ntier-\u003Ctier>.md`, or for a fork, `Workflow: \u003Cpath> (forked from \u003Cshipped tier\nfile> @ \u003Cstamped version>)`. Execute the loaded file's phase plan; no phase\ncontent is duplicated in this file.\n\n## Worker assignment (shared across tiers)\n\nRoles are indexed in [`workers.md`](..\u002Fhex-core\u002Freferences\u002Fworkers.md);\nthe orchestrator loads a full persona (spawn-prompt template, output\ncontract) only for roles in the resolved spawn set\n([spawn-selection precedence](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#spawn-selection-precedence)).\nThe model class for each role × tier is in\n[`models.md`](..\u002Fhex-core\u002Freferences\u002Fmodels.md). This table maps roles to\nplanning phases; the tier files set the actual counts.\n\n| Phase | Role | Count | Purpose |\n|---|---|---|---|\n| Discover | `architecture-explorer` | 0–1 | Map the current architecture, dependencies, reusable code |\n| Discover | `explorer` | 1–4 | Deep-dive each involved area |\n| Research | `researcher` | 0–1 (3 only when the user asks) | Technology \u002F patterns \u002F domain landscape |\n| Design | `architect` | 0–1 | ADR or system design — only for a new one-way-door decision no accepted ADR covers |\n| Review | `reviewer` (focus `spec`) | 1 | Plan ↔ design consistency; for a plan built from an accepted ADR, the decomposition only |\n| Review | `architect` | 0–1 | Trade-off honesty (new one-way door with no ADR, or the user asks) |\n| Review | `researcher` | 0–1 | SOTA \u002F known-pitfall gap check (same condition) |\n| Adversary | configured adversary skill (`plan-artifact`) | 0–1 | Cross-model review (same condition) |\n\nA project's `tiers.hex-plan.\u003Ctier>.counts` can override any Count cell above\nagainst the baseline this table sets\n([`config.md` § Merge rules](..\u002Fhex-core\u002Freferences\u002Fconfig.md#merge-rules)).\n\nConcurrency cap and degraded mode:\n[`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#worker-coordination). The\nadversary skill name comes from `hex.md › Preferences`\n([adversary contract](..\u002Fhex-core\u002Freferences\u002Fadversary.md#adversary-contract));\n`codex-adversary` is only an example value.\n\n## Project rules and conventions\n\nhex never carries a hardcoded rule or subsystem table. Every phase that needs\nthe project's invariants, verification command, or artifact conventions\ndiscovers them from **project context** (the client's ambient instructions \u002F\nproject rules), cached in the Pointers section of\n`.agents\u002Fmemory\u002Fhex.md`\n([`memory.md`](..\u002Fhex-core\u002Freferences\u002Fmemory.md#the-three-sections)). \"Verify\"\nanywhere below means **run the project's documented verification**\n([`verify.md`](..\u002Fhex-core\u002Freferences\u002Fverify.md#verification)).\n\n## The plan artifact\n\n**Location.** Write the plan where the project's documented plan\u002FADR\nconventions say (discovered from project context, cached in\n`hex.md › Pointers`). When nothing is documented, use the default artifact\nhome `.agents\u002Fplans\u002F`\n([`memory.md`](..\u002Fhex-core\u002Freferences\u002Fmemory.md#location-and-resolution)).\nRecord the active-plan pointer in `hex.md › Memory`. The plan template shipped\nwith `\u002Fhex-init` is the scaffold; the project's own format wins when it has\none. A plan carrying a `Repo` column lives in the **lead only** — one\nartifact, one Status block — with its active-plan pointer recorded solely in\nthe lead's `hex.md › Memory`\n([`memory.md`](..\u002Fhex-core\u002Freferences\u002Fmemory.md#federation-satellites)).\n\n**Status block.** Every plan opens with a self-contained status block the\nexecute and review skills read and mutate — no external state file:\n\n```markdown\n## Status\n- State:   plan-approved      \u003C!-- planning → plan-approved → executing → review → done -->\n- Tier:    high\n- Tier-grammar: 5\n- Updated: 2026-07-19\n- Next:    \u002Fhex-execute \u003Cthis plan path>\n```\n\nhex-plan initializes it at `plan-approved` on handoff, writes\n`- Tier-grammar: 5` (the tier grammar the `Tier:` value was written in,\n[`protocol.md` § Tier grammar](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#tier-grammar),\n`adr_0017` C-997) into every new plan's Status block, and records the pointer\nin `hex.md › Memory`; `\u002Fhex-execute` advances `State` and `Next` as it runs.\n\n**Required content** (every tier — the tier files scale depth, not presence):\n\n- **Component contracts** — the public surface touched (types, signatures,\n  error variants) with expected behavior and edge cases, testable enough that\n  a tester could write failing tests from them without reading any code.\n  Numbered `C-001, C-002, …`; UX scenarios below are numbered\n  `S-001, S-002, …` — both are the coverage join keys checked at review and\n  convergence\n  ([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#traceability-ids)).\n- **User-experience scenarios** — action → expected outcome → error cases for\n  each user-facing behavior.\n- **Contract wave** — the stubs **plus contract tests** for every pipeline,\n  committed once before any pipeline starts. Contract-first TDD runs inside\n  each step (Stub → Specify → Implement), never as plan-level phases. A\n  one-pipeline plan has no contract wave.\n- **Pipelines** — the plan is a few **pipelines** cut along contracts, each an\n  ordered chain of **steps**; cutting rules and the table's columns are in\n  [`decompose.md`](..\u002Fhex-core\u002Freferences\u002Fdecompose.md#parallel-by-default-decomposition).\n  A Parallelization table with one row per pipeline (id, repo, scope citing\n  the C-\u002FS- IDs it covers, expected files, wave, depends-on, marks, status —\n  initialized `pending`), the steps of each pipeline as an ordered list of\n  small briefs, a wave-grouped mermaid `graph TD` as its visual index (the\n  table stays canonical), the critical path, a \"Shippable after wave: N\"\n  line (tier medium and below exempt — one pipeline), and — when fewer\n  parallel pipelines than file-disjointness allows — a one-line\n  justification\n  ([`worktree.md`](..\u002Fhex-core\u002Freferences\u002Fworktree.md#pipeline-worktree-mechanics)).\n- **Marks** — rare, set only where the author knows better than the defaults;\n  `hard` on at most one pipeline in four, `review` on a pipeline that earns\n  its own review\n  ([`decompose.md`](..\u002Fhex-core\u002Freferences\u002Fdecompose.md#parallel-by-default-decomposition),\n  [`models.md`](..\u002Fhex-core\u002Freferences\u002Fmodels.md)). Everything else runs\n  `standard`.\n- **Open questions** — unresolved ambiguities as `[NEEDS CLARIFICATION: …]`\n  markers, **hard cap 3**. More than three means the target is underspecified;\n  raise it at the gate rather than guessing.\n\n## Constraints\n\n- Every task carries **testable acceptance criteria** — no vague behaviors.\n- **Discover runs at every tier** — never assume context.\n- **Each artifact is reviewed once; research is never repeated.** A plan\n  built from an accepted ADR (or a discussion handed off with one) never\n  re-runs research or an architect on that ADR's decisions, and its single\n  reviewer checks only the decomposition — contracts testable, pipelines cut\n  cleanly along contracts, steps sized for a fresh agent\n  ([plan-artifact scope](..\u002Fhex-core\u002Freferences\u002Floop.md#the-review-fix-loop)).\n- **Never skip Review**; never exceed the concurrency cap\n  ([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#worker-coordination)).\n- **No mid-flow questions** — ambiguity is resolved at the single gate.\n- **Persist substantial research** (more than a paragraph) as a research\n  artifact in the convention-resolved location.\n- **Always** include component contracts, UX scenarios with error cases, and\n  the Parallelization section.\n- **Always** initialize the plan's Status block and record the active-plan\n  pointer in `hex.md › Memory`.\n- **Constitution gate** — when project context names a constitution (cached\n  in `hex.md › Pointers`), the Design and Review phases gate the plan\n  against it\n  ([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#constitution-gate));\n  violations need a Constitution Deviations row.\n- **Never commit and never push** — this skill plans only.\n- **Upkeep** ([`protocol.md`](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#upkeep-step)):\n  as the final phase, re-point any `hex.md › Pointers` entry this run\n  revealed as drifted (a changed verification command, a new artifact home)\n  and update `hex.md › Memory`; a perspective gap belongs in\n  `hex.md › Memory` as a note for the next `\u002Fhex-init` run, never written to\n  `hex.md › Preferences` directly.\n\n## Handoff\n\nThe [handoff contract](..\u002Fhex-core\u002Freferences\u002Fprotocol.md#handoff-contract)\napplies: this block is the run's required final message; one optional\nproceed question may follow it.\n\n```markdown\n## Plan Complete: \u003Cfeature | \"Resolves #N\">\n\n### Classification\n- Scope: small | medium | large\n- Reversibility: two-way | one-way (medium) | one-way (high)\n- Tier: low | medium | high | xhigh | max\n- Overlays: architect=\u003Cinline|on>, research=\u003Cskip|1|3>, adversary=\u003Con|off>\n\n### Artifacts\n- \u003Cplan path> (Status block initialized; active-plan pointer in hex.md › Memory)\n- \u003Cresearch artifact path(s)>\n- \u003CADR path> (one-way-door only)\n\n### Pipelines (for \u002Fhex-execute)\n- Contract wave: stubs and contract tests to commit first (none for one pipeline)\n- Parallelization: \u003CN> pipelines, \u003CM> steps in \u003CW> waves; critical path \u003Cpipeline a → pipeline b>; marks: \u003CK> `hard`, \u003CJ> `review`\n\n### Deferred findings (need human judgment)\n- Review panel: …\n- Cross-model review: …\n\n### Next step\n    \u002Fhex-execute \u003Cplan path>\n```\n\nConsumers: `\u002Fhex-execute` (the plan artifact); the human (deferred findings).\n\n$ARGUMENTS\n",{"name":5,"license":14,"metadata":23,"description":7},{"summary":24,"keywords":25,"repository":26},"Tiered swarm planning — prompt\u002Fissue to reviewed contract-first plan","planning,swarm,multi-agent,decomposition,research,adr,tdd","https:\u002F\u002Fgithub.com\u002Fmichael-herwig\u002Farcana","main","ec018c25128e4b5b2e08679288e9e9153085830a",".claude\u002Fskills\u002Fhex-plan",1,"https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire","https:\u002F\u002Fgrimoire.rs","https:\u002F\u002Fgithub.com\u002Fgrimoire-rs","2026-11-14 09:47:24",{"source":8,"skillName":5,"commands":36,"prompt":46,"zipAvailable":47},[37,40,43],{"tool":38,"command":39},"npx","npx skills add grimoire-rs\u002Fgrimoire --skill hex-plan",{"tool":41,"command":42},"bunx","bunx skills add grimoire-rs\u002Fgrimoire --skill hex-plan",{"tool":44,"command":45},"pnpm","pnpm dlx skills add grimoire-rs\u002Fgrimoire --skill hex-plan","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhex-plan\u002FSKILL.md",true,[49,55,59,63,67,71,75],{"path":50,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":53,"sourceHtmlUrl":54},"classify.md","REFERENCE","参考文档",5513,"https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhex-plan\u002Fclassify.md",{"path":56,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":57,"sourceHtmlUrl":58},"overlays.md",5524,"https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhex-plan\u002Foverlays.md",{"path":60,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":61,"sourceHtmlUrl":62},"tier-high.md",8858,"https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhex-plan\u002Ftier-high.md",{"path":64,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":65,"sourceHtmlUrl":66},"tier-low.md",2459,"https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhex-plan\u002Ftier-low.md",{"path":68,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":69,"sourceHtmlUrl":70},"tier-max.md",2649,"https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhex-plan\u002Ftier-max.md",{"path":72,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":73,"sourceHtmlUrl":74},"tier-medium.md",4943,"https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhex-plan\u002Ftier-medium.md",{"path":76,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":77,"sourceHtmlUrl":78},"tier-xhigh.md",8560,"https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhex-plan\u002Ftier-xhigh.md",[80,88,95,102,109,116],{"skillNo":81,"name":82,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":87},"01M78R7R0RQWY8BC0AQSB3E6NX","rca","Use when a bug ticket, GitHub issue or error report needs a root cause analysis before code is touched — \"is this a bug\", \"why does this fail\". Read-only. Gives a verdict (bug or not-a-bug kind), file:line evidence, ruled-out hypotheses, other affected sites, fix direction and size. A ticket, not a diff (hex-review).","https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Frca\u002FSKILL.md",[],[],"2026-11-15 14:19:17",{"skillNo":89,"name":90,"nameZh":6,"description":91,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":92,"categories":93,"keywords":94,"publishedAt":87},"01M78R7JY3CY2XDNTB330GXTJK","fix","Use when a bug ticket or GitHub issue should go from report to an approved fix plan — \"plan a fix\", \"fix this issue\". Root-causes it via rca, then plans in plan mode with the failing regression test first. Routes to hex-plan instead when the fix is large or moves an interface contract.","https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Ffix\u002FSKILL.md",[],[],{"skillNo":96,"name":97,"nameZh":6,"description":98,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":99,"categories":100,"keywords":101,"publishedAt":18},"01M72XYVY9FGB38MZDJ0QFFMCC","p9-skill","Principle 9 differential fixture skill (v2).","https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002Ftest\u002Ftools\u002Fp9_fixtures\u002Fv2\u002Fskills\u002Fp9-skill\u002FSKILL.md",[],[],{"skillNo":103,"name":104,"nameZh":6,"description":105,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":106,"categories":107,"keywords":108,"publishedAt":18},"01M72XYT3PPTXEFWCR8H4GWKEG","p9-member","Bundle member skill for the Principle 9 fixture (v2).","https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002Ftest\u002Ftools\u002Fp9_fixtures\u002Fv2\u002Fskills\u002Fp9-member\u002FSKILL.md",[],[],{"skillNo":110,"name":111,"nameZh":6,"description":112,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":113,"categories":114,"keywords":115,"publishedAt":18},"01M72XYRWQWZ4BD0367QWG96TB","p9-hookish","Hook-free skill carrying an ignored hook.toml (v2).","https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002Ftest\u002Ftools\u002Fp9_fixtures\u002Fv2\u002Fskills\u002Fp9-hookish\u002FSKILL.md",[],[],{"skillNo":117,"name":97,"nameZh":6,"description":118,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":119,"categories":120,"keywords":121,"publishedAt":18},"01M72XYPDXXK5C6ZDD5BS2N6HS","Principle 9 differential fixture skill (v1).","https:\u002F\u002Fgithub.com\u002Fgrimoire-rs\u002Fgrimoire\u002Fblob\u002Fmain\u002Ftest\u002Ftools\u002Fp9_fixtures\u002Fv1\u002Fskills\u002Fp9-skill\u002FSKILL.md",[],[]]