[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M74ZFHA3E5JD1JM1TNF3QZKB":3,"skill-same:01M74ZFHA3E5JD1JM1TNF3QZKB":51},{"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":30,"commitSha":31,"skillDirectory":32,"forksCount":33,"repositoryHtmlUrl":34,"repositoryHomepageUrl":35,"ownerHtmlUrl":36,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":37,"install":38,"resources":50},"01M74ZFHA3E5JD1JM1TNF3QZKB","asset-placement-gate",null,"Routes a proposed skill, plugin, or agent to its correct home — forge-harness (FH) meta-skill, project-local agent, or drop — by applying a 4-criteria meta-skill bar followed by a project-local value test.","chrono-meta\u002Fforge-harness","chrono-meta","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F151837060?v=4","01M74ZDXK87ES4B30NKYSEH3MS","01M74ZDXM0F1NMYK7BS7PYR54G",33,"MIT","https:\u002F\u002Fgithub.com\u002Fchrono-meta\u002Fforge-harness\u002Fblob\u002Fmain\u002Fplugins\u002Ffh-meta\u002Fskills\u002Fasset-placement-gate\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: asset-placement-gate\ndescription: Routes a proposed skill, plugin, or agent to its correct home — forge-harness (FH) meta-skill, project-local agent, or drop — by applying a 4-criteria meta-skill bar followed by a project-local value test.\nuser-invocable: true\nallowed-tools: [\"Read\", \"Grep\", \"Glob\", \"Bash\"]\nmodel: sonnet\n---\n\n# asset-placement-gate\n\nRoutes new skills, agents, and plugins to the correct location when proposed.\n\n## Triggers\n- \"Should I make this an FH skill?\"\n- \"Where should I put this agent?\"\n- When a new FH asset creation is proposed\n- `\u002Fasset-placement-gate {asset name or description}`\n\n### Natural Language Triggers (example user phrases)\n\nWhen unsure where to place a new asset or skill:\n\n| Example phrase | Intent |\n|---|---|\n| \"Should I put this in a separate file?\" | Asset necessity + placement decision |\n| \"Can this pattern be shared across projects?\" | FH cross-project value assessment |\n| \"Is this only for our project, or can others use it too?\" | cross-project vs local routing |\n| \"What if I turn this into a skill and use it across projects?\" | FH 4-criteria trigger |\n| \"Where should I put this agent if I want to share it?\" | Placement routing needed |\n| \"Should I extract this to a separate file, or leave it?\" | Asset necessity + placement decision |\n| \"I'd like to manage this as a shared resource\" | Cross-project shared management review |\n| \"Decide whether this stays local or is available to other teams too\" | cross-project vs local routing |\n| \"I don't know where to save this\" | Placement routing needed |\n| \"Should this become a shared asset?\" | FH 4-criteria trigger |\n\n### Execution Order (summary)\n\n1. Request full file path from user (or accept natural language description)\n2. Load asset content via `Read` (if path provided)\n2.5. Step 0.5 mechanical overlap pre-scan (grounds criterion ④) + Step 0.6 official-corpora check (grounds criterion ③)\n3. Evaluate Step 1 4-criteria in order (LLM makes the judgment, ④ gated on the Step 0.5 scan, ③ informed by Step 0.6)\n4. ① + ④ both pass + at least one of ②③ passes → output **\"FH suitable\"**\n   Otherwise, proceed to Step 2 local assessment → if fails, output **\"Project-local agent or no asset needed\"**\n\n---\n\n## Step 0. Parse Input + Fetch Asset\n\nImmediately after trigger, acquire asset content in the following order.\n\n1. **Path provided** (contains `.md` or starts with `\u002F`): Read the file.\n   - If read fails: \"File not found. Please verify the path or provide a direct description.\" — then stop.\n2. **Description provided** (natural language): Check whether all 3 fields can be identified:\n   - **Purpose**: What this asset does (1 line)\n   - **Trigger**: In what situation is it invoked?\n   - **Expected callers**: Which project \u002F which users will use it?\n\n   All 3 identifiable → use as asset description.\n   Any 1 unclear → stop with this question:\n   > **The asset description is insufficient. Please provide:**\n   > 1. Purpose of this asset (one line)\n   > 2. In what situation is it invoked?\n   > 3. Which project\u002Fusers will primarily use it?\n3. **No input**: Stop with the question below.\n   > **Which asset should I evaluate?**\n   > Enter a file path (e.g., `.claude\u002Fagents\u002Fjira-create.md`) or a description.\n\nAfter acquiring the asset content, run Step 0.5 (mechanical overlap pre-scan) and Step 0.6 (official-corpora check) **before** the judged Step 1.\n\n## Step 0.5. Mechanical Overlap Pre-Scan (grounds criterion ④)\n\nCriterion ④ (\"no overlap with existing FH skills\") is otherwise an LLM **recall** judgment with no\nground truth — a duplicate skill with a novel name passes because the judge has no enumerated list to\ncheck against (judge-robustness swarm, 2026-06-13). Ground it mechanically first:\n\nRun this in **Bash** from the FH repo root (the pipeline below is `grep | grep -v | grep -c`, which the\nGrep tool cannot express).\n\n```bash\n# LEG A — enumerate existing skill names + descriptions (grounds the judged comparison).\n# This leg is ALSO the liveness check for leg B: a dead scan and a genuine no-collision both\n# produce collision-count 0, so the enumerate count must be read FIRST.\nROSTER=\"$(grep -riE 'name:|description:' plugins\u002Ffh-meta\u002Fskills\u002F*\u002FSKILL.md plugins\u002Ffh-commons\u002Fskills\u002F*\u002FSKILL.md)\"\nROSTER_N=\"$(printf '%s' \"$ROSTER\" | grep -c . )\"\necho \"roster_entries=$ROSTER_N\"\nif [ \"$ROSTER_N\" -eq 0 ]; then\n  echo \"SCAN_DEAD — enumerate leg returned 0 rows (wrong cwd \u002F missing plugins tree \u002F glob did not match).\"\n  echo \"Criterion ④ = UNDETERMINED. Do NOT read this as 'no collision'. Fix the cwd and re-run.\"\n  # fail-closed: stop here, do not run leg B, do not pass ④\nfi\n# LEG B — hard-collision check (run ONLY when ROSTER_N > 0).\n# WHOLE proposed name or a WHOLE trigger phrase reused verbatim.\n# grep -wF (whole-word, fixed-string) on the full strings — NOT -E on tokens (a shared common\n# word like \"review\" is not a collision). Exclude the asset's own file (self-match = false hit).\nSELF=\"plugins\u002Ffh-meta\u002Fskills\u002FPROPOSED_NAME\u002FSKILL.md\"\ngrep -rwF -e \"PROPOSED_FULL_NAME\" -e \"FULL_TRIGGER_PHRASE_1\" -e \"FULL_TRIGGER_PHRASE_2\" \\\n     plugins\u002F*\u002Fskills\u002F*\u002FSKILL.md | grep -v \"$SELF\" | grep -c .\n```\n\n(Replace the `PROPOSED_*` \u002F `FULL_TRIGGER_*` tokens with the literal strings under evaluation.)\n\nSurface **roster_entries + collision count + nearest existing skill(s)** — the roster count is part of\nthe report, not a private step: `collision=0` is only meaningful next to a non-zero `roster_entries`\n(not-found ≠ zero). Criterion ④ then passes only if **`roster_entries` > 0 AND 0 whole-name\u002F\nwhole-trigger collision AND the judged ≤90%-overlap check agrees**. `roster_entries = 0` is\n`SCAN_DEAD` → ④ **UNDETERMINED** (fail-closed), never a pass. A verbatim whole-name or\nwhole-trigger reuse is a hard ④ fail regardless of the LLM judgment. **Honest scope**: the grep grounds\n*literal* name\u002Ftrigger reuse only — a post-cutoff duplicate with a *paraphrased* trigger is invisible to\nboth the judge (cutoff) and the grep (literal); that residual leans on the judged half **fed the\nenumerated descriptions above** (grounded comparison, not pure memory), not on full mechanization. A\nshared common word is a judged-review flag, **not** a hard fail.\n\n**External grounding (2026-09-14 frontier signal, citation verified live before inclusion)**: Saha,\nFaghih & Feizi, *\"Under the Hood of SKILL.md: Semantic Supply-chain Attacks on AI Agent Skill\nRegistry\"* (arXiv:2605.11418, May 2026 — https:\u002F\u002Farxiv.org\u002Fabs\u002F2605.11418). Using real registry\nmechanisms (not a hypothetical), it measures SKILL.md-only attacks that manipulate which skill gets\n**discovered** (adversarial phrasing wins embedding-retrieval visibility: up to 86% pairwise win rate,\n80% Top-10 placement) and **selected** (description-only framing biases the agent toward a\nfunctionally-equivalent adversarial variant in 77.6% of paired trials on average). This is external\nconfirmation that the paraphrase\u002Fsemantic gap named just above — the grep grounds *literal* reuse\nonly — sits on a measured admission-time attack surface, not a theoretical edge case. **This note does\nnot close the residual**: no new mechanized check is added here, and none should be added by an\nautonomous pass — a `steel-quench` Wave scoped to adversarial (not merely accidental) trigger\u002Fdescription\nframing is a candidate follow-up, but the gate-design decision belongs to human review\n(§Mechanization Boundary — judgment stays with evolution).\n\n🟥 **The paper's fourth stage is the one aimed at THIS gate, and the paragraph above left it out**\n(caught 2026-09-14 by reading the abstract directly rather than re-reading the citing paragraph — the\nomission is invisible to a re-read, which is the whole argument for opening the source). Verbatim:\n*\"In Governance, semantic evasion strategies cause malicious skills to avoid a blocking verdict in\n**36.5%-100%** of cases.\"* Discovery and selection describe **which skill gets picked**; governance\ndescribes **how a blocking verdict gets dodged** — and an admission gate is exactly a blocking-verdict\nsurface, so that band is a measurement of this gate's own evasion ceiling, not of a neighbouring\nconcern. Read it against what Step 0.5 actually holds: the grep leg is literal, so semantic evasion\npasses it by construction, and the judged leg is a model reading descriptions — the same modality the\n36.5%-100% was measured on. **The honest reading is narrow in both directions**: that band is a\ndifferent registry, a different threat model (deliberately adversarial submissions, where this gate's\ncommon case is accidental duplication), and an upper end of 100% means *some* configuration blocked\nnothing at all, not that this gate does. What it does establish is that \"semantic paraphrase evades a\nliteral check\" is measured, not speculative, on a governance surface. Still no mechanized check is\nadded — for the reason the previous paragraph gives, unchanged.\n\n## Step 0.6. Tier-0\u002F1 Official-Corpora Check (grounds criterion ③)\n\nBesides the FH roster (Step 0.5), check the proposal against the official corpora: platform\nbuilt-ins, `claude-plugins-official`, and the **Claude Cookbook pattern list**\n(platform.claude.com\u002Fcookbook — agent patterns · tools · RAG · evals · skills). A hit here is a\n**judged ③ flag, not an automatic ③ fail**: it obliges the proposal to state its governance\nincrement over the official pattern (no-reinvention rule — an official pattern that fully covers\nthe need outranks a net-new build; a pattern the proposal governs *on top of* does not fail ③).\nOffline fallback: note `cookbook: unchecked` **in the Step 3 routing output** rather than silently\nskipping. (Provenance: `tracks\u002F_audit\u002Fsession_2026_07_25_claude5-context-rules-sister.md`.)\n\n## Done When\n\nEach condition declares its check class (mandatory-pass \u002F measured \u002F judged); every judged condition\nnames its adversarial pairing — no judge-only path.\n\n```\nAll steps 0–3 completed\n  (mandatory-pass — each step's output block is present; a skipped step is a FAIL, not a default pass)\n+ Step 0.5 scan is LIVE: roster_entries > 0\n  (measured: the reported roster_entries count. 0 = SCAN_DEAD → criterion ④ UNDETERMINED and this\n   skill is NOT done — a dead scan and a genuine no-collision both read 0 collisions, so the\n   liveness number is what separates them)\n+ Step 3 routing result output (location: FH meta-skill \u002F local agent \u002F drop)\n  (judged — adversarial pairing: `fh-meta:challenger` re-argues the case for the destination that was\n   NOT chosen, citing the 4 criteria; a routing verdict that survives the opposite case passes, an\n   unopposed one does not)\n+ Next action specified (write SKILL.md \u002F create .claude\u002Fagents\u002F \u002F none)\n  (mandatory-pass — a literal next-action string from that enum; blank or \"TBD\" does not satisfy it)\n```\n\n---\n\n## Step 1. FH Meta-Skill 4-Criteria Evaluation\n\n| # | Criterion | Evaluation Question |\n|:-:|---|---|\n| ① | Cross-project value | Is this asset equally useful in other projects without depending on a specific project? |\n| ② | Orchestration \u002F judgment layer | Is it just a list of MCP\u002FBash calls, or a judgment layer that synthesizes multiple signals? |\n| ③ | Not replaceable by built-ins | Can this be equally achieved with direct MCP calls or basic bash? (If yes, fails this criterion.) A Step 0.6 official-pattern hit is a judged flag: fails ③ only if the pattern fully covers the need with no governance increment |\n| ④ | No overlap with existing FH skills | Step 0.5 mechanical scan = 0 name\u002Ftrigger collision **AND** judged ≤90% overlap. Non-zero collision → hard fail. |\n\n**FH suitable** → ① + ④ both pass + at least one of ②③ passes.\n**Fail** → ① or ④ fails → immediate fail. Or both ②③ fail → proceed to Step 2.\n\n---\n\n## Step 2. Project-Local Agent Value Assessment\n\nFor assets that failed Step 1:\n\n| # | Criterion | Evaluation Question |\n|:-:|---|---|\n| A | Project-specific knowledge | Does it encode paths, conventions, or domain rules specific to a project? |\n| B | Repeated workflow pattern | Is it a workflow performed repeatedly within that project? |\n| C | Differentiation from built-ins | Does it provide local-context-based convenience such as automatic convention application or step integration? |\n\n**Local agent suitable** → A satisfied + at least one of B·C.\n→ Recommend creating `{project}\u002F.claude\u002Fagents\u002F{name}.md`.\n\n**Drop** → A not satisfied or both B·C not satisfied.\n→ Equivalent to built-in capability. No asset needed.\n\n---\n\n## Step 3. Routing Result Output\n\n```\n[asset-placement-gate verdict]\n\nAsset: {asset name}\nDescription: {one-line summary}\n\n── FH 4-criteria ──────────────────\n① Cross-project value     : O \u002F X\n② Orchestration\u002Fjudgment  : O \u002F X\n③ Not replaceable         : O \u002F X\n④ No overlap with existing : O \u002F X  ← required gate with ① (immediate fail if either fails)\n→ FH suitable: Pass \u002F Fail  (① + ④ required + at least one of ②③)\n\n── Project-local assessment ──────  (if FH failed)\nA. Project-specific knowledge : O \u002F X\nB. Repeated workflow pattern  : O \u002F X\nC. Differentiation from built-ins : O \u002F X\n→ Local agent: Suitable \u002F Drop\n\n── Conclusion ────────────────────\nLocation: [FH meta-skill | {project} local agent | Drop]\nNext action: [Write SKILL.md | Create .claude\u002Fagents\u002F | None]\n```\n","# asset-placement-gate\n\nRoutes new skills, agents, and plugins to the correct location when proposed.\n\n## Triggers\n- \"Should I make this an FH skill?\"\n- \"Where should I put this agent?\"\n- When a new FH asset creation is proposed\n- `\u002Fasset-placement-gate {asset name or description}`\n\n### Natural Language Triggers (example user phrases)\n\nWhen unsure where to place a new asset or skill:\n\n| Example phrase | Intent |\n|---|---|\n| \"Should I put this in a separate file?\" | Asset necessity + placement decision |\n| \"Can this pattern be shared across projects?\" | FH cross-project value assessment |\n| \"Is this only for our project, or can others use it too?\" | cross-project vs local routing |\n| \"What if I turn this into a skill and use it across projects?\" | FH 4-criteria trigger |\n| \"Where should I put this agent if I want to share it?\" | Placement routing needed |\n| \"Should I extract this to a separate file, or leave it?\" | Asset necessity + placement decision |\n| \"I'd like to manage this as a shared resource\" | Cross-project shared management review |\n| \"Decide whether this stays local or is available to other teams too\" | cross-project vs local routing |\n| \"I don't know where to save this\" | Placement routing needed |\n| \"Should this become a shared asset?\" | FH 4-criteria trigger |\n\n### Execution Order (summary)\n\n1. Request full file path from user (or accept natural language description)\n2. Load asset content via `Read` (if path provided)\n2.5. Step 0.5 mechanical overlap pre-scan (grounds criterion ④) + Step 0.6 official-corpora check (grounds criterion ③)\n3. Evaluate Step 1 4-criteria in order (LLM makes the judgment, ④ gated on the Step 0.5 scan, ③ informed by Step 0.6)\n4. ① + ④ both pass + at least one of ②③ passes → output **\"FH suitable\"**\n   Otherwise, proceed to Step 2 local assessment → if fails, output **\"Project-local agent or no asset needed\"**\n\n---\n\n## Step 0. Parse Input + Fetch Asset\n\nImmediately after trigger, acquire asset content in the following order.\n\n1. **Path provided** (contains `.md` or starts with `\u002F`): Read the file.\n   - If read fails: \"File not found. Please verify the path or provide a direct description.\" — then stop.\n2. **Description provided** (natural language): Check whether all 3 fields can be identified:\n   - **Purpose**: What this asset does (1 line)\n   - **Trigger**: In what situation is it invoked?\n   - **Expected callers**: Which project \u002F which users will use it?\n\n   All 3 identifiable → use as asset description.\n   Any 1 unclear → stop with this question:\n   > **The asset description is insufficient. Please provide:**\n   > 1. Purpose of this asset (one line)\n   > 2. In what situation is it invoked?\n   > 3. Which project\u002Fusers will primarily use it?\n3. **No input**: Stop with the question below.\n   > **Which asset should I evaluate?**\n   > Enter a file path (e.g., `.claude\u002Fagents\u002Fjira-create.md`) or a description.\n\nAfter acquiring the asset content, run Step 0.5 (mechanical overlap pre-scan) and Step 0.6 (official-corpora check) **before** the judged Step 1.\n\n## Step 0.5. Mechanical Overlap Pre-Scan (grounds criterion ④)\n\nCriterion ④ (\"no overlap with existing FH skills\") is otherwise an LLM **recall** judgment with no\nground truth — a duplicate skill with a novel name passes because the judge has no enumerated list to\ncheck against (judge-robustness swarm, 2026-06-13). Ground it mechanically first:\n\nRun this in **Bash** from the FH repo root (the pipeline below is `grep | grep -v | grep -c`, which the\nGrep tool cannot express).\n\n```bash\n# LEG A — enumerate existing skill names + descriptions (grounds the judged comparison).\n# This leg is ALSO the liveness check for leg B: a dead scan and a genuine no-collision both\n# produce collision-count 0, so the enumerate count must be read FIRST.\nROSTER=\"$(grep -riE 'name:|description:' plugins\u002Ffh-meta\u002Fskills\u002F*\u002FSKILL.md plugins\u002Ffh-commons\u002Fskills\u002F*\u002FSKILL.md)\"\nROSTER_N=\"$(printf '%s' \"$ROSTER\" | grep -c . )\"\necho \"roster_entries=$ROSTER_N\"\nif [ \"$ROSTER_N\" -eq 0 ]; then\n  echo \"SCAN_DEAD — enumerate leg returned 0 rows (wrong cwd \u002F missing plugins tree \u002F glob did not match).\"\n  echo \"Criterion ④ = UNDETERMINED. Do NOT read this as 'no collision'. Fix the cwd and re-run.\"\n  # fail-closed: stop here, do not run leg B, do not pass ④\nfi\n# LEG B — hard-collision check (run ONLY when ROSTER_N > 0).\n# WHOLE proposed name or a WHOLE trigger phrase reused verbatim.\n# grep -wF (whole-word, fixed-string) on the full strings — NOT -E on tokens (a shared common\n# word like \"review\" is not a collision). Exclude the asset's own file (self-match = false hit).\nSELF=\"plugins\u002Ffh-meta\u002Fskills\u002FPROPOSED_NAME\u002FSKILL.md\"\ngrep -rwF -e \"PROPOSED_FULL_NAME\" -e \"FULL_TRIGGER_PHRASE_1\" -e \"FULL_TRIGGER_PHRASE_2\" \\\n     plugins\u002F*\u002Fskills\u002F*\u002FSKILL.md | grep -v \"$SELF\" | grep -c .\n```\n\n(Replace the `PROPOSED_*` \u002F `FULL_TRIGGER_*` tokens with the literal strings under evaluation.)\n\nSurface **roster_entries + collision count + nearest existing skill(s)** — the roster count is part of\nthe report, not a private step: `collision=0` is only meaningful next to a non-zero `roster_entries`\n(not-found ≠ zero). Criterion ④ then passes only if **`roster_entries` > 0 AND 0 whole-name\u002F\nwhole-trigger collision AND the judged ≤90%-overlap check agrees**. `roster_entries = 0` is\n`SCAN_DEAD` → ④ **UNDETERMINED** (fail-closed), never a pass. A verbatim whole-name or\nwhole-trigger reuse is a hard ④ fail regardless of the LLM judgment. **Honest scope**: the grep grounds\n*literal* name\u002Ftrigger reuse only — a post-cutoff duplicate with a *paraphrased* trigger is invisible to\nboth the judge (cutoff) and the grep (literal); that residual leans on the judged half **fed the\nenumerated descriptions above** (grounded comparison, not pure memory), not on full mechanization. A\nshared common word is a judged-review flag, **not** a hard fail.\n\n**External grounding (2026-09-14 frontier signal, citation verified live before inclusion)**: Saha,\nFaghih & Feizi, *\"Under the Hood of SKILL.md: Semantic Supply-chain Attacks on AI Agent Skill\nRegistry\"* (arXiv:2605.11418, May 2026 — https:\u002F\u002Farxiv.org\u002Fabs\u002F2605.11418). Using real registry\nmechanisms (not a hypothetical), it measures SKILL.md-only attacks that manipulate which skill gets\n**discovered** (adversarial phrasing wins embedding-retrieval visibility: up to 86% pairwise win rate,\n80% Top-10 placement) and **selected** (description-only framing biases the agent toward a\nfunctionally-equivalent adversarial variant in 77.6% of paired trials on average). This is external\nconfirmation that the paraphrase\u002Fsemantic gap named just above — the grep grounds *literal* reuse\nonly — sits on a measured admission-time attack surface, not a theoretical edge case. **This note does\nnot close the residual**: no new mechanized check is added here, and none should be added by an\nautonomous pass — a `steel-quench` Wave scoped to adversarial (not merely accidental) trigger\u002Fdescription\nframing is a candidate follow-up, but the gate-design decision belongs to human review\n(§Mechanization Boundary — judgment stays with evolution).\n\n🟥 **The paper's fourth stage is the one aimed at THIS gate, and the paragraph above left it out**\n(caught 2026-09-14 by reading the abstract directly rather than re-reading the citing paragraph — the\nomission is invisible to a re-read, which is the whole argument for opening the source). Verbatim:\n*\"In Governance, semantic evasion strategies cause malicious skills to avoid a blocking verdict in\n**36.5%-100%** of cases.\"* Discovery and selection describe **which skill gets picked**; governance\ndescribes **how a blocking verdict gets dodged** — and an admission gate is exactly a blocking-verdict\nsurface, so that band is a measurement of this gate's own evasion ceiling, not of a neighbouring\nconcern. Read it against what Step 0.5 actually holds: the grep leg is literal, so semantic evasion\npasses it by construction, and the judged leg is a model reading descriptions — the same modality the\n36.5%-100% was measured on. **The honest reading is narrow in both directions**: that band is a\ndifferent registry, a different threat model (deliberately adversarial submissions, where this gate's\ncommon case is accidental duplication), and an upper end of 100% means *some* configuration blocked\nnothing at all, not that this gate does. What it does establish is that \"semantic paraphrase evades a\nliteral check\" is measured, not speculative, on a governance surface. Still no mechanized check is\nadded — for the reason the previous paragraph gives, unchanged.\n\n## Step 0.6. Tier-0\u002F1 Official-Corpora Check (grounds criterion ③)\n\nBesides the FH roster (Step 0.5), check the proposal against the official corpora: platform\nbuilt-ins, `claude-plugins-official`, and the **Claude Cookbook pattern list**\n(platform.claude.com\u002Fcookbook — agent patterns · tools · RAG · evals · skills). A hit here is a\n**judged ③ flag, not an automatic ③ fail**: it obliges the proposal to state its governance\nincrement over the official pattern (no-reinvention rule — an official pattern that fully covers\nthe need outranks a net-new build; a pattern the proposal governs *on top of* does not fail ③).\nOffline fallback: note `cookbook: unchecked` **in the Step 3 routing output** rather than silently\nskipping. (Provenance: `tracks\u002F_audit\u002Fsession_2026_07_25_claude5-context-rules-sister.md`.)\n\n## Done When\n\nEach condition declares its check class (mandatory-pass \u002F measured \u002F judged); every judged condition\nnames its adversarial pairing — no judge-only path.\n\n```\nAll steps 0–3 completed\n  (mandatory-pass — each step's output block is present; a skipped step is a FAIL, not a default pass)\n+ Step 0.5 scan is LIVE: roster_entries > 0\n  (measured: the reported roster_entries count. 0 = SCAN_DEAD → criterion ④ UNDETERMINED and this\n   skill is NOT done — a dead scan and a genuine no-collision both read 0 collisions, so the\n   liveness number is what separates them)\n+ Step 3 routing result output (location: FH meta-skill \u002F local agent \u002F drop)\n  (judged — adversarial pairing: `fh-meta:challenger` re-argues the case for the destination that was\n   NOT chosen, citing the 4 criteria; a routing verdict that survives the opposite case passes, an\n   unopposed one does not)\n+ Next action specified (write SKILL.md \u002F create .claude\u002Fagents\u002F \u002F none)\n  (mandatory-pass — a literal next-action string from that enum; blank or \"TBD\" does not satisfy it)\n```\n\n---\n\n## Step 1. FH Meta-Skill 4-Criteria Evaluation\n\n| # | Criterion | Evaluation Question |\n|:-:|---|---|\n| ① | Cross-project value | Is this asset equally useful in other projects without depending on a specific project? |\n| ② | Orchestration \u002F judgment layer | Is it just a list of MCP\u002FBash calls, or a judgment layer that synthesizes multiple signals? |\n| ③ | Not replaceable by built-ins | Can this be equally achieved with direct MCP calls or basic bash? (If yes, fails this criterion.) A Step 0.6 official-pattern hit is a judged flag: fails ③ only if the pattern fully covers the need with no governance increment |\n| ④ | No overlap with existing FH skills | Step 0.5 mechanical scan = 0 name\u002Ftrigger collision **AND** judged ≤90% overlap. Non-zero collision → hard fail. |\n\n**FH suitable** → ① + ④ both pass + at least one of ②③ passes.\n**Fail** → ① or ④ fails → immediate fail. Or both ②③ fail → proceed to Step 2.\n\n---\n\n## Step 2. Project-Local Agent Value Assessment\n\nFor assets that failed Step 1:\n\n| # | Criterion | Evaluation Question |\n|:-:|---|---|\n| A | Project-specific knowledge | Does it encode paths, conventions, or domain rules specific to a project? |\n| B | Repeated workflow pattern | Is it a workflow performed repeatedly within that project? |\n| C | Differentiation from built-ins | Does it provide local-context-based convenience such as automatic convention application or step integration? |\n\n**Local agent suitable** → A satisfied + at least one of B·C.\n→ Recommend creating `{project}\u002F.claude\u002Fagents\u002F{name}.md`.\n\n**Drop** → A not satisfied or both B·C not satisfied.\n→ Equivalent to built-in capability. No asset needed.\n\n---\n\n## Step 3. Routing Result Output\n\n```\n[asset-placement-gate verdict]\n\nAsset: {asset name}\nDescription: {one-line summary}\n\n── FH 4-criteria ──────────────────\n① Cross-project value     : O \u002F X\n② Orchestration\u002Fjudgment  : O \u002F X\n③ Not replaceable         : O \u002F X\n④ No overlap with existing : O \u002F X  ← required gate with ① (immediate fail if either fails)\n→ FH suitable: Pass \u002F Fail  (① + ④ required + at least one of ②③)\n\n── Project-local assessment ──────  (if FH failed)\nA. Project-specific knowledge : O \u002F X\nB. Repeated workflow pattern  : O \u002F X\nC. Differentiation from built-ins : O \u002F X\n→ Local agent: Suitable \u002F Drop\n\n── Conclusion ────────────────────\nLocation: [FH meta-skill | {project} local agent | Drop]\nNext action: [Write SKILL.md | Create .claude\u002Fagents\u002F | None]\n```\n",{"name":5,"model":23,"description":7,"allowed-tools":24,"user-invocable":29},"sonnet",[25,26,27,28],"Read","Grep","Glob","Bash",true,"main","899495d17a23f30ca149d4ac29a137a8fa7864e3","plugins\u002Ffh-meta\u002Fskills\u002Fasset-placement-gate",1,"https:\u002F\u002Fgithub.com\u002Fchrono-meta\u002Fforge-harness","https:\u002F\u002Fchrono-meta.github.io\u002Fforge-harness\u002F","https:\u002F\u002Fgithub.com\u002Fchrono-meta","2026-11-15 02:42:21",{"source":8,"skillName":5,"commands":39,"prompt":49,"zipAvailable":29},[40,43,46],{"tool":41,"command":42},"npx","npx skills add chrono-meta\u002Fforge-harness --skill asset-placement-gate",{"tool":44,"command":45},"bunx","bunx skills add chrono-meta\u002Fforge-harness --skill asset-placement-gate",{"tool":47,"command":48},"pnpm","pnpm dlx skills add chrono-meta\u002Fforge-harness --skill asset-placement-gate","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fchrono-meta\u002Fforge-harness\u002Fblob\u002Fmain\u002Fplugins\u002Ffh-meta\u002Fskills\u002Fasset-placement-gate\u002FSKILL.md",[],[52,59,66,73,79,86],{"skillNo":53,"name":54,"nameZh":6,"description":55,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":56,"categories":57,"keywords":58,"publishedAt":18},"01M74ZPW36H1HVE0R61QZ5ZND0","todo-orchestrator","TODO CLI 프로젝트의 기능 추가·버그 수정·리팩터링·테스트 보강을 개발자와 QA 에이전트로 조율한다. todo 앱 관련 코드 작업 요청에 사용한다. 다시 실행, 재실행, 수정, 보완, 이전 결과 개선, QA만 다시 같은 후속 요청에도 사용한다. 단순 질문에는 사용하지 않는다.","https:\u002F\u002Fgithub.com\u002Fchrono-meta\u002Fforge-harness\u002Fblob\u002Fmain\u002Fscripts\u002Ffixtures\u002Fgenerated_harness\u002FP4_r2\u002F.claude\u002Fskills\u002Ftodo-orchestrator\u002FSKILL.md",[],[],{"skillNo":60,"name":61,"nameZh":6,"description":62,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":63,"categories":64,"keywords":65,"publishedAt":18},"01M74ZPS4EP2349B346R5JQEYE","todo-dev","TODO CLI(todo\u002Fcli.py)에 명령을 추가하거나 버그를 고치고 테스트를 작성하는 방법. todo 앱 코드 수정·기능 추가·테스트 작성·리팩터링 작업에 사용한다. 수정, 보완, 다시 구현 요청에도 사용한다.","https:\u002F\u002Fgithub.com\u002Fchrono-meta\u002Fforge-harness\u002Fblob\u002Fmain\u002Fscripts\u002Ffixtures\u002Fgenerated_harness\u002FP4_r2\u002F.claude\u002Fskills\u002Ftodo-dev\u002FSKILL.md",[],[],{"skillNo":67,"name":68,"nameZh":6,"description":69,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":70,"categories":71,"keywords":72,"publishedAt":18},"01M74ZPPX2743J6FPT8N2ZT0D5","todo-dev-team","TODO CLI(todo\u002Fcli.py) 개발 작업을 개발자·테스터·리뷰어 팀으로 진행하는 오케스트레이터. 명령 추가, 버그 수정, 테스트 작성, 리팩터링, 커밋·푸시 요청에 사용한다. 다시 실행, 재실행, 수정, 보완, 리뷰 반영, 테스트만 다시 같은 후속 요청에도 사용한다. 단순 질문에는 쓰지 않는다.","https:\u002F\u002Fgithub.com\u002Fchrono-meta\u002Fforge-harness\u002Fblob\u002Fmain\u002Fscripts\u002Ffixtures\u002Fgenerated_harness\u002FP3_r2\u002F.claude\u002Fskills\u002Ftodo-dev-team\u002FSKILL.md",[],[],{"skillNo":74,"name":61,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":18},"01M74ZPKWST7DF24H0EN9RCEME","TODO CLI 개발 작업(명령·기능 추가, 버그 수정, 테스트 작성, 리뷰, 커밋·push)을 개발자·테스터·리뷰어 에이전트 팀으로 진행하는 오케스트레이터. todo CLI, todo.json, todo\u002Fcli.py 관련 구현·수정·테스트 요청에 사용한다. 재실행, 수정, 보완, 이전 결과 개선, 테스트만 다시, 리뷰만 다시 요청에도 사용한다. 단순 질문에는 쓰지 않는다.","https:\u002F\u002Fgithub.com\u002Fchrono-meta\u002Fforge-harness\u002Fblob\u002Fmain\u002Fscripts\u002Ffixtures\u002Fgenerated_harness\u002FP3_r1\u002F.claude\u002Fskills\u002Ftodo-dev\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":18},"01M74ZPHBP62X3KKSV2RVKSBKT","weekly-summary-write","정리된 회의록으로 주간 회의 요약 문서를 작성하는 형식과 문체 규칙. 주간 요약 작성, 요약 문서 수정·보완, 검토 의견 반영 시 사용한다.","https:\u002F\u002Fgithub.com\u002Fchrono-meta\u002Fforge-harness\u002Fblob\u002Fmain\u002Fscripts\u002Ffixtures\u002Fgenerated_harness\u002FP2_r2\u002F.claude\u002Fskills\u002Fweekly-summary-write\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":18},"01M74ZPECSXY4C0QNHAD96Z88H","weekly-meeting-summary","주간 회의록을 정리하고 요약 문서를 만드는 전체 작업을 조율하는 오케스트레이터. 회의록 요약, 주간 회의 정리, 주간 보고용 요약 요청 시 사용한다. 다시 실행, 재실행, 요약 수정·보완, 이전 결과를 바탕으로 한 개선, 정리만\u002F검토만 다시 같은 후속 요청에도 사용한다.","https:\u002F\u002Fgithub.com\u002Fchrono-meta\u002Fforge-harness\u002Fblob\u002Fmain\u002Fscripts\u002Ffixtures\u002Fgenerated_harness\u002FP2_r2\u002F.claude\u002Fskills\u002Fweekly-meeting-summary\u002FSKILL.md",[],[]]