[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M73FNE7J4P9MHT0Z7WSBNVFZ":3,"skill-same:01M73FNE7J4P9MHT0Z7WSBNVFZ":63},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"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":41,"commitSha":42,"skillDirectory":43,"forksCount":44,"repositoryHtmlUrl":45,"repositoryHomepageUrl":6,"ownerHtmlUrl":46,"ownerBlogUrl":6,"ownerDescription":47,"updatedAt":48,"install":49,"resources":62},"01M73FNE7J4P9MHT0Z7WSBNVFZ","project-review",null,"Performs a comprehensive review of the current project: machine validator baseline first, then parallel specialist agents (scope-triageable: full \u002F scoped \u002F baseline-only). Auto-detects project type and agent roster, produces a prioritized, class-tagged findings report persisted to docs\u002Freports\u002F, and wires outcomes to the ticket backlog (deferred items, validator-hardening loop). Use when: user requests a full project review (\"\u002Fproject-review\" or \"do a full project review\"); PM detects structural changes (3+ agent files modified, phase schema changes, workspace-schema.json modified, new variant added); QA escalation from auditor (audit.ts ERROR >= 3 or security Critical finding).","5throck\u002Fai-workspace-standards","5throck","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F10958964?v=4","01M73BFEH0V0VS483TW4JD49JW","01M73BFEJ5VQZJHJSFTVPAZGDM",25,"AGPL-3.0","https:\u002F\u002Fgithub.com\u002F5throck\u002Fai-workspace-standards\u002Fblob\u002Fmain\u002Ftemplates\u002Fco-safety\u002F.agents\u002Fskills\u002Fproject-review\u002FSKILL.md",[],[],"2026-11-11 15:11:14",false,"---\nname: project-review\nstatus: active\nscope: co-safety\ndescription: >\n  Performs a comprehensive review of the current project: machine validator\n  baseline first, then parallel specialist agents (scope-triageable: full \u002F\n  scoped \u002F baseline-only). Auto-detects project type and agent roster,\n  produces a prioritized, class-tagged findings report persisted to\n  docs\u002Freports\u002F, and wires outcomes to the ticket backlog (deferred items,\n  validator-hardening loop).\n  Use when: user requests a full project review (\"\u002Fproject-review\" or\n  \"do a full project review\"); PM detects structural changes (3+ agent files modified,\n  phase schema changes, workspace-schema.json modified, new variant added);\n  QA escalation from auditor (audit.ts ERROR >= 3 or security Critical finding).\nowner: pm\nversion: 1.3.4\nlast_reviewed: 2026-10-11\nprerequisites: []\nmetadata:\n  type: process\n  triggers:\n    - project review\n    - review project\n    - audit project\n    - quality review\n  related_skills:\n    - project-resync\n    - upgrade-project\n    - meeting-facilitation\n---\n\n# project-review\n\n## Context\n\nUse this skill to perform project review with project-local validation commands and evidence capture.\n\nComprehensive review of the current project: machine baseline → scope-triaged specialist\nreview → persisted report → ticketed follow-ups. **Ratchet principle**: every finding\nclassified `script-gap` must end as a validator-hardening ticket so the next review\ncatches that class mechanically, not by agent effort.\n\n## When to Use\n\n- User explicitly requests a full project review\n- PM detects structural changes requiring cross-domain validation (T-02)\n- QA escalation: `audit.ts` exits with 3+ ERRORs, or security-expert finds a Critical issue\n- Post-incident localized review (use `scoped` mode — see Step 1.5)\n- Weekly health pulse (use `baseline-only` mode — doubles as the §9.1 check)\n\n## Documentation\u002Flink ownership\n\n`project-review` owns routine documentation link-check evidence. Use `baseline-only` mode for weekly or pre-closeout docs\u002Flink validation.\n\n## Step 0 — Machine Baseline (run BEFORE any agent dispatch)\n\nRun the validator battery and record results — this is (a) the report's Baseline\nsection and (b) the reference for classifying findings as `script-gap` later.\n\nOne-shot consolidated runner (T-20260912-030): `bun scripts\u002Freview-baseline.ts` is the workspace-root equivalent —\nruns the battery below in read-only mode with a PASS\u002FFAIL summary\n(`--quiet` for summary only). In this project, run the project-local commands:\n\n```bash\nbun run audit                                     # project\u002Fworkspace standards\nbun run verify-scripts --verify                   # SCRIPTS.md registry sync\nbun run agent-lifecycle-audit                     # agent health\nbun run skill-lifecycle-audit                     # skill health\n\nWhen operating in the workspace root, also run the template integrity and L1\u002FL2 drift gates documented in the root maintainer playbook.\n```\n\nRecord per script: PASS\u002FFAIL\u002FWARN counts. In variant projects, run the project's own\n`bun scripts\u002Faudit.ts` instead of the root battery.\n\nIf the baseline already shows ≥3 ERRORs, triage those FIRST (T-03) — agents may still be\ndispatched for non-machine-detectable classes, but don't let agents re-discover what the\nscripts just printed.\n\n### base-map MCP (optional enhancement)\n\nCheck whether `mcp__base-map__*` tools exist in the session. If yes, set\n`BASE_MAP_AVAILABLE = true` and use them for large-file summarization,\ncross-validation of findings, and `review_code` on critical scripts. If no,\nskip every `mcp__base-map__*` call — the review completes without them.\nbase-map is an enhancer, never a dependency.\n\n## Step 1 — Detect Project Context\n\n1. **List available agents**: scan `agents\u002F` for `*.md` (excluding README)\n2. **Determine project type**: `docs\u002Fcontext.md` present → variant project; workspace\n   root indicators otherwise\n3. **Announce context**:\n   ```\n   Project type: [workspace-root | co-… ]\n   Available agents: [list]\n   Review domains: [mapped]\n   Scope: [full | scoped:\u003Cdomains> | baseline-only]\n   base-map MCP: [available | not available]\n   ```\n\n## Step 1.5 — Scope Triage (choose review breadth)\n\n| Mode | Agents | When |\n|------|--------|------|\n| `full` | 7 domains, parallel (paired per Step 3) | Default for T-01 user request, T-02 structural change, T-03 QA escalation |\n| `scoped \u003Cdomains>` | Only the named domains | Post-incident review where the blast radius is known (2026-09-07 lesson: a localized registry-tag mismatch did not need 7 agents) |\n| `baseline-only` | Zero — machine battery + PM lightweight check (§7 universal behaviors spot-check) | Weekly health pulse; quick pre-release sanity; when agent budget is constrained |\n\nSelection rule: the PM picks the *smallest* mode that covers the trigger's blast radius,\nand states the choice in the Step 1 announcement.\n\n## Step 2 — Generate Execution Plan\n\nMap available agents to review domains (workspace root shown; variant projects map to\ntheir own roster, PM covers gaps directly):\n\n| # | Domain | Agent | Tier | Focus |\n|---|--------|-------|------|-------|\n| 1 | Architecture + Scaffolding | architect (else PM) | High | Structure, variant contracts, template sync, L0\u002FL1\u002FL2 inheritance |\n| 2 | Standards + Lifecycle | auditor (else PM) | Medium | Registry\u002Fmanifest accuracy, lifecycle records, governance doc consistency |\n| 3 | Automation | automation-engineer (else PM) | Medium | Scripts, hooks, package.json, CI workflows, cross-platform |\n| 4 | Documentation + Security | docs-writer + security-expert (else PM) | Medium | Links, language policy, secrets, CI permissions |\n\n> If an agent is not available for a domain, PM covers that domain directly with a\n> lightweight check. For `scoped` mode, keep only the selected rows.\n\n## Step 3 — Dispatch Agents (resilient parallel)\n\n**Cap: at most 4 background agents.** Domains are pre-paired to fit the cap — this is\nthe combination proven in the 2026-09-08 run:\n\n| Slot | Domains |\n|------|---------|\n| Agent A | Architecture + Scaffolding |\n| Agent B | Standards + Lifecycle |\n| Agent C | Automation |\n| Agent D | Documentation + Security |\n\nDispatch all slots in ONE message via the `Agent` tool with `run_in_background: true`\n(`subagent_type: Explore` for read-only reviews). Prompt template:\n\n```\nYou are the [domain(s)] reviewer for this project at [path].\nReview your domain and report: Critical, High, Moderate issues and Strengths.\nEvery finding MUST include: file path (+line number where possible), evidence\nyou actually verified (diff\u002Fread output, not assumption), and a suggested fix.\nResearch only — do NOT modify any files.\n[If BASE_MAP_AVAILABLE: use mcp__base-map__ask_local_llm to summarize large\nfiles and cross-validate; mcp__base-map__review_code on critical scripts.]\n```\n\n**Fallbacks**:\n- Background dispatch blocked by concurrency limits → run slots sequentially\n  (foreground), or merge two slots into one prompt.\n- On Antigravity\u002FGemini CLI: the meeting-facilitation skill with topic \"project review\" and options `--agents [list] --rounds 2 --dialogue` (the legacy `\u002Fmeeting` slash command is retired).\n- No Agent tool at all: PM role-plays each slot sequentially using the same prompts.\n\n## Step 4 — Collect, Classify, Persist\n\n### 4a — Collect raw findings\nGather per-slot reports: 🔴 Critical \u002F 🟡 High \u002F 🟢 Moderate \u002F ✅ Strengths.\n\n### 4b — Cross-domain dedup\nMerge duplicates (same root cause), credit all discoverers, keep the most specific\nfile:line and the highest severity.\n\n### 4c — base-map cross-validation (optional, only if `BASE_MAP_AVAILABLE`)\n`mcp__base-map__ask_local_llm` over the merged findings for false positives,\nmissing issues, and root-cause chains.\n\n### 4d — Findings table (Class column is MANDATORY)\n\n```markdown\n## Review Results — [Project] — [Date]\n\n**Baseline**: audit PASS · validate-templates 0\u002F0 · verify-scripts N PASS · drift: gemini-settings only\n\n### 🔴 Critical (fix immediately)\n| # | Issue | Agent | File:Line | Class | Fix |\n|---|-------|-------|-----------|-------|-----|\n\n### 🟡 High (fix within 1 week)   \u003C!-- same columns -->\n### 🟢 Moderate (fix within 2 weeks) \u003C!-- same columns -->\n### ℹ️ Low \u002F Improvements\n### ✅ Strengths\n```\n\n**Class values** (drives Step 5 wiring):\n- `one-time` — instance drift; fix and forget (e.g. a stale version field)\n- `systemic` — recurring by nature but needs judgment each time (e.g. boilerplate\n  replication across 13 variants)\n- `script-gap` — a *machine* could and should have caught this; nobody did\n  (e.g. the 2026-09-08 L1 workspace-schema drift → became the WS-01 parity check)\n\n### 4e — Persist the report\n\nWrite to `docs\u002Freports\u002FYYYY-MM-DD-project-review-\u003Cscope>.md` following the directory's\nheader convention:\n\n```markdown\n# Project Review — [Project] — [Date]\n**Date**: [YYYY-MM-DD]\n**Scope**: [workspace root | variant | domains list]\n**Method**: [4 parallel agents | scoped | baseline-only] + machine battery\n\u003C!-- If no fixes were applied in-session: -->\n> Analysis only — no files modified in this report.\n```\n\nInclude: baseline, findings tables (4d), domain summary, action wiring results (Step 5),\nand — after fixes — the verification results (Step 6).\n\n## Step 5 — Wire Outcomes to Action\n\nRoute findings by Class — do not default to \"fix everything now\".\n\n| Route | Condition | Action |\n|-------|-----------|--------|\n| Fix now | Critical\u002FHigh, fix is clear, session budget allows | Dispatch through the normal PM Gateway (specialists execute) |\n| Ticket | Deferred items (needs design gate, low priority, future cycle) | In workspace-root reviews, create a governance ticket with the ticket helper; in variant projects, record the deferred item in the report and memory log. |\n| Validator-hardening ticket | Any `script-gap` finding | Same `create --manual`, title prefixed `validator-hardening:` — tracks the ratchet loop until a standing check exists |\n\nRules:\n- Ticket titles in English, one actionable sentence.\n- Record ticket IDs back into the persisted report's \"Action wiring\" section.\n- The ticket helper is workspace-root only (L0); in variant projects, record\n  deferred items in the report + memory log instead.\n\n## Step 6 — Post-Fix Verification\n\nIf fixes were applied this session:\n\n1. Re-run every validator whose domain was touched (minimum: `bun scripts\u002Faudit.ts`).\n2. Append a `## Verification` section to the persisted report with the results.\n3. Log the run in `memory\u002FYYYY-MM-DD.md` per the daily-log convention (summary, changes,\n   deferred\u002Fticket list, validation line).\n\n**Ratchet check (quarterly, AGENTS.md §10)**: diff persisted `docs\u002Freports\u002F*project-review*` files —\n`script-gap` classes should move to the machine baseline over time (found-by-agent ↓,\ncaught-by-script ↑). If the same class reappears as an agent finding, the hardening\nticket was not landed.\n\n## Trigger Reference\n\n| Trigger | Invoker | Condition | Default scope |\n|---------|---------|-----------|---------------|\n| T-01: User request | User | `\u002Fproject-review` or natural language equivalent | full (scoped on request) |\n| T-02: PM autonomous | PM | 3+ agent files modified; phase schema changed; workspace-schema.json modified; new variant added | full |\n| T-03: QA escalation | auditor \u002F security | audit.ts ERROR ≥ 3; security Critical | scoped to the failing domain, then full if it spreads |\n\n## Related Skills\n\n- **project-resync**: fleet-level close-out after fixes land (commit\u002FPR pipeline)\n- **upgrade-project**: delivers template updates to the project\n- Loop: project-review (diagnose) → project-resync (fleet sync\u002Fbackport) → upgrade-project (deliver) → project-review baseline-only (verify).\n- **meeting-facilitation**: Antigravity\u002FGemini dispatch path for Step 3\n\n## Output Format\n\nProduce a prioritized findings report.\n\n## Execution Steps\n\nFollow the baseline, specialist review, synthesis, and verification steps below.\n","# project-review\n\n## Context\n\nUse this skill to perform project review with project-local validation commands and evidence capture.\n\nComprehensive review of the current project: machine baseline → scope-triaged specialist\nreview → persisted report → ticketed follow-ups. **Ratchet principle**: every finding\nclassified `script-gap` must end as a validator-hardening ticket so the next review\ncatches that class mechanically, not by agent effort.\n\n## When to Use\n\n- User explicitly requests a full project review\n- PM detects structural changes requiring cross-domain validation (T-02)\n- QA escalation: `audit.ts` exits with 3+ ERRORs, or security-expert finds a Critical issue\n- Post-incident localized review (use `scoped` mode — see Step 1.5)\n- Weekly health pulse (use `baseline-only` mode — doubles as the §9.1 check)\n\n## Documentation\u002Flink ownership\n\n`project-review` owns routine documentation link-check evidence. Use `baseline-only` mode for weekly or pre-closeout docs\u002Flink validation.\n\n## Step 0 — Machine Baseline (run BEFORE any agent dispatch)\n\nRun the validator battery and record results — this is (a) the report's Baseline\nsection and (b) the reference for classifying findings as `script-gap` later.\n\nOne-shot consolidated runner (T-20260912-030): `bun scripts\u002Freview-baseline.ts` is the workspace-root equivalent —\nruns the battery below in read-only mode with a PASS\u002FFAIL summary\n(`--quiet` for summary only). In this project, run the project-local commands:\n\n```bash\nbun run audit                                     # project\u002Fworkspace standards\nbun run verify-scripts --verify                   # SCRIPTS.md registry sync\nbun run agent-lifecycle-audit                     # agent health\nbun run skill-lifecycle-audit                     # skill health\n\nWhen operating in the workspace root, also run the template integrity and L1\u002FL2 drift gates documented in the root maintainer playbook.\n```\n\nRecord per script: PASS\u002FFAIL\u002FWARN counts. In variant projects, run the project's own\n`bun scripts\u002Faudit.ts` instead of the root battery.\n\nIf the baseline already shows ≥3 ERRORs, triage those FIRST (T-03) — agents may still be\ndispatched for non-machine-detectable classes, but don't let agents re-discover what the\nscripts just printed.\n\n### base-map MCP (optional enhancement)\n\nCheck whether `mcp__base-map__*` tools exist in the session. If yes, set\n`BASE_MAP_AVAILABLE = true` and use them for large-file summarization,\ncross-validation of findings, and `review_code` on critical scripts. If no,\nskip every `mcp__base-map__*` call — the review completes without them.\nbase-map is an enhancer, never a dependency.\n\n## Step 1 — Detect Project Context\n\n1. **List available agents**: scan `agents\u002F` for `*.md` (excluding README)\n2. **Determine project type**: `docs\u002Fcontext.md` present → variant project; workspace\n   root indicators otherwise\n3. **Announce context**:\n   ```\n   Project type: [workspace-root | co-… ]\n   Available agents: [list]\n   Review domains: [mapped]\n   Scope: [full | scoped:\u003Cdomains> | baseline-only]\n   base-map MCP: [available | not available]\n   ```\n\n## Step 1.5 — Scope Triage (choose review breadth)\n\n| Mode | Agents | When |\n|------|--------|------|\n| `full` | 7 domains, parallel (paired per Step 3) | Default for T-01 user request, T-02 structural change, T-03 QA escalation |\n| `scoped \u003Cdomains>` | Only the named domains | Post-incident review where the blast radius is known (2026-09-07 lesson: a localized registry-tag mismatch did not need 7 agents) |\n| `baseline-only` | Zero — machine battery + PM lightweight check (§7 universal behaviors spot-check) | Weekly health pulse; quick pre-release sanity; when agent budget is constrained |\n\nSelection rule: the PM picks the *smallest* mode that covers the trigger's blast radius,\nand states the choice in the Step 1 announcement.\n\n## Step 2 — Generate Execution Plan\n\nMap available agents to review domains (workspace root shown; variant projects map to\ntheir own roster, PM covers gaps directly):\n\n| # | Domain | Agent | Tier | Focus |\n|---|--------|-------|------|-------|\n| 1 | Architecture + Scaffolding | architect (else PM) | High | Structure, variant contracts, template sync, L0\u002FL1\u002FL2 inheritance |\n| 2 | Standards + Lifecycle | auditor (else PM) | Medium | Registry\u002Fmanifest accuracy, lifecycle records, governance doc consistency |\n| 3 | Automation | automation-engineer (else PM) | Medium | Scripts, hooks, package.json, CI workflows, cross-platform |\n| 4 | Documentation + Security | docs-writer + security-expert (else PM) | Medium | Links, language policy, secrets, CI permissions |\n\n> If an agent is not available for a domain, PM covers that domain directly with a\n> lightweight check. For `scoped` mode, keep only the selected rows.\n\n## Step 3 — Dispatch Agents (resilient parallel)\n\n**Cap: at most 4 background agents.** Domains are pre-paired to fit the cap — this is\nthe combination proven in the 2026-09-08 run:\n\n| Slot | Domains |\n|------|---------|\n| Agent A | Architecture + Scaffolding |\n| Agent B | Standards + Lifecycle |\n| Agent C | Automation |\n| Agent D | Documentation + Security |\n\nDispatch all slots in ONE message via the `Agent` tool with `run_in_background: true`\n(`subagent_type: Explore` for read-only reviews). Prompt template:\n\n```\nYou are the [domain(s)] reviewer for this project at [path].\nReview your domain and report: Critical, High, Moderate issues and Strengths.\nEvery finding MUST include: file path (+line number where possible), evidence\nyou actually verified (diff\u002Fread output, not assumption), and a suggested fix.\nResearch only — do NOT modify any files.\n[If BASE_MAP_AVAILABLE: use mcp__base-map__ask_local_llm to summarize large\nfiles and cross-validate; mcp__base-map__review_code on critical scripts.]\n```\n\n**Fallbacks**:\n- Background dispatch blocked by concurrency limits → run slots sequentially\n  (foreground), or merge two slots into one prompt.\n- On Antigravity\u002FGemini CLI: the meeting-facilitation skill with topic \"project review\" and options `--agents [list] --rounds 2 --dialogue` (the legacy `\u002Fmeeting` slash command is retired).\n- No Agent tool at all: PM role-plays each slot sequentially using the same prompts.\n\n## Step 4 — Collect, Classify, Persist\n\n### 4a — Collect raw findings\nGather per-slot reports: 🔴 Critical \u002F 🟡 High \u002F 🟢 Moderate \u002F ✅ Strengths.\n\n### 4b — Cross-domain dedup\nMerge duplicates (same root cause), credit all discoverers, keep the most specific\nfile:line and the highest severity.\n\n### 4c — base-map cross-validation (optional, only if `BASE_MAP_AVAILABLE`)\n`mcp__base-map__ask_local_llm` over the merged findings for false positives,\nmissing issues, and root-cause chains.\n\n### 4d — Findings table (Class column is MANDATORY)\n\n```markdown\n## Review Results — [Project] — [Date]\n\n**Baseline**: audit PASS · validate-templates 0\u002F0 · verify-scripts N PASS · drift: gemini-settings only\n\n### 🔴 Critical (fix immediately)\n| # | Issue | Agent | File:Line | Class | Fix |\n|---|-------|-------|-----------|-------|-----|\n\n### 🟡 High (fix within 1 week)   \u003C!-- same columns -->\n### 🟢 Moderate (fix within 2 weeks) \u003C!-- same columns -->\n### ℹ️ Low \u002F Improvements\n### ✅ Strengths\n```\n\n**Class values** (drives Step 5 wiring):\n- `one-time` — instance drift; fix and forget (e.g. a stale version field)\n- `systemic` — recurring by nature but needs judgment each time (e.g. boilerplate\n  replication across 13 variants)\n- `script-gap` — a *machine* could and should have caught this; nobody did\n  (e.g. the 2026-09-08 L1 workspace-schema drift → became the WS-01 parity check)\n\n### 4e — Persist the report\n\nWrite to `docs\u002Freports\u002FYYYY-MM-DD-project-review-\u003Cscope>.md` following the directory's\nheader convention:\n\n```markdown\n# Project Review — [Project] — [Date]\n**Date**: [YYYY-MM-DD]\n**Scope**: [workspace root | variant | domains list]\n**Method**: [4 parallel agents | scoped | baseline-only] + machine battery\n\u003C!-- If no fixes were applied in-session: -->\n> Analysis only — no files modified in this report.\n```\n\nInclude: baseline, findings tables (4d), domain summary, action wiring results (Step 5),\nand — after fixes — the verification results (Step 6).\n\n## Step 5 — Wire Outcomes to Action\n\nRoute findings by Class — do not default to \"fix everything now\".\n\n| Route | Condition | Action |\n|-------|-----------|--------|\n| Fix now | Critical\u002FHigh, fix is clear, session budget allows | Dispatch through the normal PM Gateway (specialists execute) |\n| Ticket | Deferred items (needs design gate, low priority, future cycle) | In workspace-root reviews, create a governance ticket with the ticket helper; in variant projects, record the deferred item in the report and memory log. |\n| Validator-hardening ticket | Any `script-gap` finding | Same `create --manual`, title prefixed `validator-hardening:` — tracks the ratchet loop until a standing check exists |\n\nRules:\n- Ticket titles in English, one actionable sentence.\n- Record ticket IDs back into the persisted report's \"Action wiring\" section.\n- The ticket helper is workspace-root only (L0); in variant projects, record\n  deferred items in the report + memory log instead.\n\n## Step 6 — Post-Fix Verification\n\nIf fixes were applied this session:\n\n1. Re-run every validator whose domain was touched (minimum: `bun scripts\u002Faudit.ts`).\n2. Append a `## Verification` section to the persisted report with the results.\n3. Log the run in `memory\u002FYYYY-MM-DD.md` per the daily-log convention (summary, changes,\n   deferred\u002Fticket list, validation line).\n\n**Ratchet check (quarterly, AGENTS.md §10)**: diff persisted `docs\u002Freports\u002F*project-review*` files —\n`script-gap` classes should move to the machine baseline over time (found-by-agent ↓,\ncaught-by-script ↑). If the same class reappears as an agent finding, the hardening\nticket was not landed.\n\n## Trigger Reference\n\n| Trigger | Invoker | Condition | Default scope |\n|---------|---------|-----------|---------------|\n| T-01: User request | User | `\u002Fproject-review` or natural language equivalent | full (scoped on request) |\n| T-02: PM autonomous | PM | 3+ agent files modified; phase schema changed; workspace-schema.json modified; new variant added | full |\n| T-03: QA escalation | auditor \u002F security | audit.ts ERROR ≥ 3; security Critical | scoped to the failing domain, then full if it spreads |\n\n## Related Skills\n\n- **project-resync**: fleet-level close-out after fixes land (commit\u002FPR pipeline)\n- **upgrade-project**: delivers template updates to the project\n- Loop: project-review (diagnose) → project-resync (fleet sync\u002Fbackport) → upgrade-project (deliver) → project-review baseline-only (verify).\n- **meeting-facilitation**: Antigravity\u002FGemini dispatch path for Step 3\n\n## Output Format\n\nProduce a prioritized findings report.\n\n## Execution Steps\n\nFollow the baseline, specialist review, synthesis, and verification steps below.\n",{"name":5,"owner":23,"scope":24,"status":25,"version":26,"metadata":27,"description":38,"last_reviewed":39,"prerequisites":40},"pm","co-safety","active","1.3.4",{"type":28,"triggers":29,"related_skills":34},"process",[30,31,32,33],"project review","review project","audit project","quality review",[35,36,37],"project-resync","upgrade-project","meeting-facilitation","Performs a comprehensive review of the current project: machine validator baseline first, then parallel specialist agents (scope-triageable: full \u002F scoped \u002F baseline-only). Auto-detects project type and agent roster, produces a prioritized, class-tagged findings report persisted to docs\u002Freports\u002F, and wires outcomes to the ticket backlog (deferred items, validator-hardening loop). Use when: user requests a full project review (\"\u002Fproject-review\" or \"do a full project review\"); PM detects structural changes (3+ agent files modified, phase schema changes, workspace-schema.json modified, new variant added); QA escalation from auditor (audit.ts ERROR >= 3 or security Critical finding).\n",1791676800000,[],"main","1162ccc568cf90c01b3886d6143cbfd851a37883","templates\u002Fco-safety\u002F.agents\u002Fskills\u002Fproject-review",9,"https:\u002F\u002Fgithub.com\u002F5throck\u002Fai-workspace-standards","https:\u002F\u002Fgithub.com\u002F5throck","Experienced in digital transformation, enterprise IT, startup acceleration, and business strategy. Focused on building practical systems that connect technolo","2026-11-14 21:22:34",{"source":8,"skillName":5,"commands":50,"prompt":60,"zipAvailable":61},[51,54,57],{"tool":52,"command":53},"npx","npx skills add 5throck\u002Fai-workspace-standards --skill project-review",{"tool":55,"command":56},"bunx","bunx skills add 5throck\u002Fai-workspace-standards --skill project-review",{"tool":58,"command":59},"pnpm","pnpm dlx skills add 5throck\u002Fai-workspace-standards --skill project-review","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002F5throck\u002Fai-workspace-standards\u002Fblob\u002Fmain\u002Ftemplates\u002Fco-safety\u002F.agents\u002Fskills\u002Fproject-review\u002FSKILL.md",true,[],[64,72,77,82,87,92],{"skillNo":65,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":68,"categories":69,"keywords":70,"publishedAt":71},"01M7BBJDYHJMGTJYSC37JVWY8S","i18n-layout","Text layout and encoding guidance: character encoding (UTF-8, legacy Korean code pages, BOM hazards), line endings, RTL\u002Fbidi handling, and script-specific font selection (Hangul focus). Use when: encoding is corrupted or mojibake appears, an RTL locale must render correctly, fonts must be chosen for a script, or CRLF\u002FBOM issues surface.","https:\u002F\u002Fgithub.com\u002F5throck\u002Fai-workspace-standards\u002Fblob\u002Fmain\u002Ftemplates\u002Fco-work\u002Fskills\u002Fi18n-layout\u002FSKILL.md",[],[],"2026-11-15 14:19:17",{"skillNo":73,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":71},"01M7BBJ9XHAAS1AVGKZZ841VBZ","https:\u002F\u002Fgithub.com\u002F5throck\u002Fai-workspace-standards\u002Fblob\u002Fmain\u002Ftemplates\u002Fco-work\u002F.hermes\u002Fskills\u002Fi18n-layout\u002FSKILL.md",[],[],{"skillNo":78,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":71},"01M7BBJ5C3X8TKFKG5R9P9P49W","https:\u002F\u002Fgithub.com\u002F5throck\u002Fai-workspace-standards\u002Fblob\u002Fmain\u002Ftemplates\u002Fco-work\u002F.gemini\u002Fskills\u002Fi18n-layout\u002FSKILL.md",[],[],{"skillNo":83,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":71},"01M7BBJ1B52X0QMJ2QVBMPJB37","https:\u002F\u002Fgithub.com\u002F5throck\u002Fai-workspace-standards\u002Fblob\u002Fmain\u002Ftemplates\u002Fco-work\u002F.codex\u002Fskills\u002Fi18n-layout\u002FSKILL.md",[],[],{"skillNo":88,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":71},"01M7BBHX91BW2Q2R276MRX5AW3","https:\u002F\u002Fgithub.com\u002F5throck\u002Fai-workspace-standards\u002Fblob\u002Fmain\u002Ftemplates\u002Fco-work\u002F.claude\u002Fskills\u002Fi18n-layout\u002FSKILL.md",[],[],{"skillNo":93,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":94,"categories":95,"keywords":96,"publishedAt":71},"01M7BBHQWGXPSFRWN1PRC2NMMV","https:\u002F\u002Fgithub.com\u002F5throck\u002Fai-workspace-standards\u002Fblob\u002Fmain\u002Ftemplates\u002Fco-work\u002F.agents\u002Fskills\u002Fi18n-layout\u002FSKILL.md",[],[]]