[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7ASNB0EBZNX4A1934YEY3P2":3,"skill-same:01M7ASNB0EBZNX4A1934YEY3P2":52},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":25,"commitSha":26,"skillDirectory":27,"forksCount":28,"repositoryHtmlUrl":29,"repositoryHomepageUrl":6,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":31,"updatedAt":32,"install":33,"resources":45},"01M7ASNB0EBZNX4A1934YEY3P2","plugin-curation",null,"Use when the user wants to audit or prune their Claude Code setup — says \"audit my plugins\", \"what am I actually using\", \"curate my Claude setup\", \"which plugins can I disable\", \"clean up my enabled plugins\" — or runs a periodic (e.g. quarterly) review of installed plugins, skills, and MCP servers. Produces keep \u002F consolidate \u002F prune recommendations grounded in real usage counts.","ozzycodes2\u002Fclaudellum","ozzycodes2","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F21285094?v=4","01M7ASJ57KTP519F96S4QGM7C5","Osniel Quintana","01M7ASJ589JS8S075AA92Y08S4",1,"MIT","https:\u002F\u002Fgithub.com\u002Fozzycodes2\u002Fclaudellum\u002Fblob\u002Fmain\u002Fsession-improv\u002Fskills\u002Fplugin-curation\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: plugin-curation\ndescription: Use when the user wants to audit or prune their Claude Code setup — says \"audit my plugins\", \"what am I actually using\", \"curate my Claude setup\", \"which plugins can I disable\", \"clean up my enabled plugins\" — or runs a periodic (e.g. quarterly) review of installed plugins, skills, and MCP servers. Produces keep \u002F consolidate \u002F prune recommendations grounded in real usage counts.\ndisable-model-invocation: true\n---\n\n# Plugin Curation\n\nAn audit pass over an installed Claude Code setup: pull real usage signals, cross-reference what's enabled, produce keep \u002F consolidate \u002F prune recommendations. The output decides nothing on its own — it surfaces evidence so the user can prune with confidence.\n\n**Core principle:** disable on evidence, not vibes — and read the evidence honestly. A low invocation count is a *question*, not a verdict: the count is blind to hooks, MCP servers, LSPs, subagents, and passive reference skills. **A favorited plugin is never an automatic cut.**\n\n## When to use\n\n- User suspects their plugin\u002Fskill\u002FMCP set has grown into bloat.\n- A periodic checkpoint (quarterly fits the data window most setups retain).\n- Before sharing a setup, or when context\u002Fstartup feels heavier than the work justifies.\n\n**Skip if:** user just wants to toggle one known plugin — that's a `\u002Fplugin` action, not an audit.\n\n## How to run\n\nWork is: gather signals (steps 1–3), categorize (step 4), apply the caveat (step 5), then recommend (step 6). **This skill never edits `settings.json`** — it recommends; user disables via `\u002Fplugin` or `update-config` afterward.\n\n### 1. Read the primary signals — `~\u002F.claude.json`\n\nHolds authoritative all-time counters. Read three keys (ignore a fourth):\n\n```bash\npython3 - \u003C\u003C 'PY'\nimport json, time, os\nd = json.load(open(os.path.expanduser(\"~\u002F.claude.json\")))\nnow = time.time()\nprint(\"=== skillUsage: count | days_since_last | skill ===\")\nfor k, v in sorted(d.get(\"skillUsage\", {}).items(), key=lambda x: -x[1][\"usageCount\"]):\n    age = round((now - v[\"lastUsedAt\"]\u002F1000)\u002F86400, 1)\n    print(f\"{v['usageCount']:>3}  {age:>6}d  {k}\")\nprint(\"=== favoritePlugins (explicit keeps) ===\")\nprint(\"\\n\".join(d.get(\"favoritePlugins\", [])))\nprint(\"=== mcpServers ===\")\nprint(\", \".join((d.get(\"mcpServers\") or {}).keys()))\nPY\n```\n\n- **`skillUsage`** — `usageCount` *and* `lastUsedAt`. Recency is half the signal: a high count that's 90 days stale means *abandoned*, not *valued* (classic case: a skill you migrated away from but never disabled). Rank by count, but read the age column next to it.\n- **`favoritePlugins`** — user's explicit pins. Treat every entry as a keep. If usage data argues for cutting one, surface the conflict — don't override the pin.\n- **`mcpServers`** — user-configured MCP servers. Their value lives in `mcp__*` calls (step 2), never in a Skill count.\n- **`toolUsage`** — present but unreliable (typically near-empty). Don't base anything on it.\n\n### 2. Fill the gaps — cc-query\n\n`skillUsage` only covers skills. For MCP traffic and subagents — where whole plugins earn their place invisibly — run the batched supplement:\n\n```bash\nCCQ=$(ls ~\u002F.claude\u002Fplugins\u002Fcache\u002Fcc-query-dev\u002Fcc-query\u002F*\u002Fbin\u002Fcc-query | head -1)\n\"$CCQ\" \u003C \u003Cpath-to>\u002Fqueries.sql      # the queries.sql sitting next to this SKILL.md\n```\n\n`queries.sql` (this skill's directory) returns MCP calls by server, subagent spawns by type (spawn tool is `Agent`, not `Task`), and the data window. Note the earliest timestamp — it bounds every count.\n\n### 3. Read what's enabled and what's orphaned\n\n```bash\n# Enabled (true) \u002F disabled (false) — the source of truth for \"what loads\"\npython3 -c \"import json,os;d=json.load(open(os.path.expanduser('~\u002F.claude\u002Fsettings.json')));[print(('ON ' if v else 'OFF'),k) for k,v in sorted(d['enabledPlugins'].items())]\"\n\n# Dead cache orphans: installed but marked orphaned (don't load)\nfd --hidden --max-depth 4 '^\\.orphaned_at$' ~\u002F.claude\u002Fplugins\u002Fcache -x dirname 2>\u002Fdev\u002Fnull\n```\n\nA plugin cached but absent from `enabledPlugins` entirely is also dead weight — corroborates the `.orphaned_at` marker.\n\n### 4. Categorize each enabled plugin\n\n| Bucket | Signal | Action |\n|---|---|---|\n| Live core | count > 0 and recently used | Keep |\n| Favorited | listed in `favoritePlugins` | Keep (flag if usage is low — user's call) |\n| Abandoned | high count but weeks\u002Fmonths stale | Question: did you migrate away? Likely disable |\n| Off-domain, zero use | no use AND irrelevant to user's actual work | Prune candidate |\n| Redundant cluster | several enabled plugins cover one job; user has a preferred one | Consolidate to the preferred; rest are candidates |\n| Dead cache orphan | `.orphaned_at` and\u002For absent from the enabled map | Already off; optional cache clear |\n| Zero-use but defensible | count 0 but provides a hook \u002F MCP \u002F LSP \u002F subagent \u002F passive reference | Keep; flag for user to confirm |\n\n### 5. Apply the caveat — *before* recommending anything\n\nA zero count does **not** mean unused. Re-check the right signal for each blind spot:\n\n- **Hooks** fire on events (SessionStart, PreToolUse, Stop) and never appear as a tool call. Check `~\u002F.claude\u002Fplugins\u002Fcache\u002F\u003Cplugin>\u002F*\u002Fhooks\u002F` and the `hooks` block in settings.\n- **MCP servers** surface only as `mcp__*` calls (step 2), never as a Skill.\n- **LSP** plugins (e.g. a language server) are consumed silently by edit\u002Fdiagnostic flows — they read as zero. Never prune an LSP on count.\n- **Subagents** — a plugin may exist only to supply an agent type; cross-reference the `Agent` spawn counts.\n- **Passive reference skills** shape behavior by being read into context, not invoked. Low count ≠ low value.\n- **Output styles \u002F status lines** have no invocation surface at all.\n\nRule of thumb: prune only the intersection of *(near-zero use) AND (not favorited) AND (no passive role) AND (off-domain or redundant)*. Anything in just one of those sets is a question for the user, not a deletion.\n\n### 6. Produce recommendations\n\nGroup as **Keep \u002F Consolidate \u002F Prune**, each entry with its evidence (count, recency, favorite flag, or the passive role that defends it). Open with the data window. Close with a single recommended first move — usually the lowest-risk, highest-noise-reduction cut. Leave the actual `settings.json` edit to the user.\n\n## Anti-patterns\n\n| Don't | Why |\n|---|---|\n| Treat a zero count as \"unused, prune it\" | Blind to hooks, MCP, LSP, subagents, and passive reference skills. Zero is a question. |\n| Recommend cutting a favorited plugin without flagging it | The pin is an explicit keep signal. Surface the count\u002Ffavorite conflict; let the user decide. |\n| Ignore recency | A 90-day-stale skill with a high count is abandoned, not valued. Cutting a freshly-used one is worse. |\n| Prune an MCP-only or LSP-only plugin on a missing Skill call | Its value lives in `mcp__*` calls or silent language-server traffic. Check that signal instead. |\n| Fire one cc-query call per metric | Per-invocation startup cost. Batch into the one `queries.sql` run. |\n| Trust `toolUsage` from `~\u002F.claude.json` | It's near-empty\u002Funreliable. Use cc-query over transcripts for tool\u002FMCP counts. |\n| Report counts with no window | \"Used 0 times\" is meaningless without \"…in N days, data back to \u003Cdate>.\" |\n| Edit `settings.json` from this skill | This is a recommendation pass. Auto-disabling on inference is how you lose a hook you needed. |\n| Score the setup with a number (\"health: 7\u002F10\") | Numbers feel rigorous and aren't. Specific keep\u002Fprune calls are the deliverable. |\n\n## Example\n\n```\nUser: audit my plugins, I think I have too many\n\nYou: [reads ~\u002F.claude.json skillUsage + favoritePlugins; runs queries.sql; reads enabledPlugins]\n\n     Data window: 2026-04-17 → today (~6 weeks, 14 projects). Counts under-represent\n     anything older than that.\n\n     Keep — live core\n       ceej-skills (session-retro 27×\u002Ftoday, writing-commit-messages 7×\u002Fthis week)\n       codex (codex-rescue: 25 Agent spawns) · trivia (458 mcp__trivia__* calls)\n       rust-analyzer-lsp — 0 Skill calls, but it's the language server. Silent, essential.\n\n     Keep — favorited (your explicit pins)\n       code-review, comprehensive-review, code-simplifier — pinned, but each used ≤2×\n       and weeks stale. Still want them handy, or has your own review skill replaced them?\n\n     Abandoned — high count, now stale\n       superpowers:brainstorming 19×, writing-plans 14× — all 28–99d stale, plugin already\n       disabled. Migration looks complete; nothing to do.\n\n     Prune candidates — off-domain, zero use, not favorited\n       observability-monitoring, database-migrations, backend-api-security — 0 use,\n       no fit with your Rust work. Safe to disable.\n\n     Recommended first move: disable the off-domain trio (zero risk, biggest noise cut).\n     Re-run next quarter — the window deepens and the picture sharpens.\n```\n","# Plugin Curation\n\nAn audit pass over an installed Claude Code setup: pull real usage signals, cross-reference what's enabled, produce keep \u002F consolidate \u002F prune recommendations. The output decides nothing on its own — it surfaces evidence so the user can prune with confidence.\n\n**Core principle:** disable on evidence, not vibes — and read the evidence honestly. A low invocation count is a *question*, not a verdict: the count is blind to hooks, MCP servers, LSPs, subagents, and passive reference skills. **A favorited plugin is never an automatic cut.**\n\n## When to use\n\n- User suspects their plugin\u002Fskill\u002FMCP set has grown into bloat.\n- A periodic checkpoint (quarterly fits the data window most setups retain).\n- Before sharing a setup, or when context\u002Fstartup feels heavier than the work justifies.\n\n**Skip if:** user just wants to toggle one known plugin — that's a `\u002Fplugin` action, not an audit.\n\n## How to run\n\nWork is: gather signals (steps 1–3), categorize (step 4), apply the caveat (step 5), then recommend (step 6). **This skill never edits `settings.json`** — it recommends; user disables via `\u002Fplugin` or `update-config` afterward.\n\n### 1. Read the primary signals — `~\u002F.claude.json`\n\nHolds authoritative all-time counters. Read three keys (ignore a fourth):\n\n```bash\npython3 - \u003C\u003C 'PY'\nimport json, time, os\nd = json.load(open(os.path.expanduser(\"~\u002F.claude.json\")))\nnow = time.time()\nprint(\"=== skillUsage: count | days_since_last | skill ===\")\nfor k, v in sorted(d.get(\"skillUsage\", {}).items(), key=lambda x: -x[1][\"usageCount\"]):\n    age = round((now - v[\"lastUsedAt\"]\u002F1000)\u002F86400, 1)\n    print(f\"{v['usageCount']:>3}  {age:>6}d  {k}\")\nprint(\"=== favoritePlugins (explicit keeps) ===\")\nprint(\"\\n\".join(d.get(\"favoritePlugins\", [])))\nprint(\"=== mcpServers ===\")\nprint(\", \".join((d.get(\"mcpServers\") or {}).keys()))\nPY\n```\n\n- **`skillUsage`** — `usageCount` *and* `lastUsedAt`. Recency is half the signal: a high count that's 90 days stale means *abandoned*, not *valued* (classic case: a skill you migrated away from but never disabled). Rank by count, but read the age column next to it.\n- **`favoritePlugins`** — user's explicit pins. Treat every entry as a keep. If usage data argues for cutting one, surface the conflict — don't override the pin.\n- **`mcpServers`** — user-configured MCP servers. Their value lives in `mcp__*` calls (step 2), never in a Skill count.\n- **`toolUsage`** — present but unreliable (typically near-empty). Don't base anything on it.\n\n### 2. Fill the gaps — cc-query\n\n`skillUsage` only covers skills. For MCP traffic and subagents — where whole plugins earn their place invisibly — run the batched supplement:\n\n```bash\nCCQ=$(ls ~\u002F.claude\u002Fplugins\u002Fcache\u002Fcc-query-dev\u002Fcc-query\u002F*\u002Fbin\u002Fcc-query | head -1)\n\"$CCQ\" \u003C \u003Cpath-to>\u002Fqueries.sql      # the queries.sql sitting next to this SKILL.md\n```\n\n`queries.sql` (this skill's directory) returns MCP calls by server, subagent spawns by type (spawn tool is `Agent`, not `Task`), and the data window. Note the earliest timestamp — it bounds every count.\n\n### 3. Read what's enabled and what's orphaned\n\n```bash\n# Enabled (true) \u002F disabled (false) — the source of truth for \"what loads\"\npython3 -c \"import json,os;d=json.load(open(os.path.expanduser('~\u002F.claude\u002Fsettings.json')));[print(('ON ' if v else 'OFF'),k) for k,v in sorted(d['enabledPlugins'].items())]\"\n\n# Dead cache orphans: installed but marked orphaned (don't load)\nfd --hidden --max-depth 4 '^\\.orphaned_at$' ~\u002F.claude\u002Fplugins\u002Fcache -x dirname 2>\u002Fdev\u002Fnull\n```\n\nA plugin cached but absent from `enabledPlugins` entirely is also dead weight — corroborates the `.orphaned_at` marker.\n\n### 4. Categorize each enabled plugin\n\n| Bucket | Signal | Action |\n|---|---|---|\n| Live core | count > 0 and recently used | Keep |\n| Favorited | listed in `favoritePlugins` | Keep (flag if usage is low — user's call) |\n| Abandoned | high count but weeks\u002Fmonths stale | Question: did you migrate away? Likely disable |\n| Off-domain, zero use | no use AND irrelevant to user's actual work | Prune candidate |\n| Redundant cluster | several enabled plugins cover one job; user has a preferred one | Consolidate to the preferred; rest are candidates |\n| Dead cache orphan | `.orphaned_at` and\u002For absent from the enabled map | Already off; optional cache clear |\n| Zero-use but defensible | count 0 but provides a hook \u002F MCP \u002F LSP \u002F subagent \u002F passive reference | Keep; flag for user to confirm |\n\n### 5. Apply the caveat — *before* recommending anything\n\nA zero count does **not** mean unused. Re-check the right signal for each blind spot:\n\n- **Hooks** fire on events (SessionStart, PreToolUse, Stop) and never appear as a tool call. Check `~\u002F.claude\u002Fplugins\u002Fcache\u002F\u003Cplugin>\u002F*\u002Fhooks\u002F` and the `hooks` block in settings.\n- **MCP servers** surface only as `mcp__*` calls (step 2), never as a Skill.\n- **LSP** plugins (e.g. a language server) are consumed silently by edit\u002Fdiagnostic flows — they read as zero. Never prune an LSP on count.\n- **Subagents** — a plugin may exist only to supply an agent type; cross-reference the `Agent` spawn counts.\n- **Passive reference skills** shape behavior by being read into context, not invoked. Low count ≠ low value.\n- **Output styles \u002F status lines** have no invocation surface at all.\n\nRule of thumb: prune only the intersection of *(near-zero use) AND (not favorited) AND (no passive role) AND (off-domain or redundant)*. Anything in just one of those sets is a question for the user, not a deletion.\n\n### 6. Produce recommendations\n\nGroup as **Keep \u002F Consolidate \u002F Prune**, each entry with its evidence (count, recency, favorite flag, or the passive role that defends it). Open with the data window. Close with a single recommended first move — usually the lowest-risk, highest-noise-reduction cut. Leave the actual `settings.json` edit to the user.\n\n## Anti-patterns\n\n| Don't | Why |\n|---|---|\n| Treat a zero count as \"unused, prune it\" | Blind to hooks, MCP, LSP, subagents, and passive reference skills. Zero is a question. |\n| Recommend cutting a favorited plugin without flagging it | The pin is an explicit keep signal. Surface the count\u002Ffavorite conflict; let the user decide. |\n| Ignore recency | A 90-day-stale skill with a high count is abandoned, not valued. Cutting a freshly-used one is worse. |\n| Prune an MCP-only or LSP-only plugin on a missing Skill call | Its value lives in `mcp__*` calls or silent language-server traffic. Check that signal instead. |\n| Fire one cc-query call per metric | Per-invocation startup cost. Batch into the one `queries.sql` run. |\n| Trust `toolUsage` from `~\u002F.claude.json` | It's near-empty\u002Funreliable. Use cc-query over transcripts for tool\u002FMCP counts. |\n| Report counts with no window | \"Used 0 times\" is meaningless without \"…in N days, data back to \u003Cdate>.\" |\n| Edit `settings.json` from this skill | This is a recommendation pass. Auto-disabling on inference is how you lose a hook you needed. |\n| Score the setup with a number (\"health: 7\u002F10\") | Numbers feel rigorous and aren't. Specific keep\u002Fprune calls are the deliverable. |\n\n## Example\n\n```\nUser: audit my plugins, I think I have too many\n\nYou: [reads ~\u002F.claude.json skillUsage + favoritePlugins; runs queries.sql; reads enabledPlugins]\n\n     Data window: 2026-04-17 → today (~6 weeks, 14 projects). Counts under-represent\n     anything older than that.\n\n     Keep — live core\n       ceej-skills (session-retro 27×\u002Ftoday, writing-commit-messages 7×\u002Fthis week)\n       codex (codex-rescue: 25 Agent spawns) · trivia (458 mcp__trivia__* calls)\n       rust-analyzer-lsp — 0 Skill calls, but it's the language server. Silent, essential.\n\n     Keep — favorited (your explicit pins)\n       code-review, comprehensive-review, code-simplifier — pinned, but each used ≤2×\n       and weeks stale. Still want them handy, or has your own review skill replaced them?\n\n     Abandoned — high count, now stale\n       superpowers:brainstorming 19×, writing-plans 14× — all 28–99d stale, plugin already\n       disabled. Migration looks complete; nothing to do.\n\n     Prune candidates — off-domain, zero use, not favorited\n       observability-monitoring, database-migrations, backend-api-security — 0 use,\n       no fit with your Rust work. Safe to disable.\n\n     Recommended first move: disable the off-domain trio (zero risk, biggest noise cut).\n     Re-run next quarter — the window deepens and the picture sharpens.\n```\n",{"name":5,"description":7,"disable-model-invocation":24},true,"main","a3f19bcc313571bff79dca391a83f9c4d38bb335","session-improv\u002Fskills\u002Fplugin-curation",0,"https:\u002F\u002Fgithub.com\u002Fozzycodes2\u002Fclaudellum","https:\u002F\u002Fgithub.com\u002Fozzycodes2","Software Engineer | Engineering Manager | AI whisperer | ☕Enjoyer","2026-11-14 04:50:11",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":24},[35,38,41],{"tool":36,"command":37},"npx","npx skills add ozzycodes2\u002Fclaudellum --skill plugin-curation",{"tool":39,"command":40},"bunx","bunx skills add ozzycodes2\u002Fclaudellum --skill plugin-curation",{"tool":42,"command":43},"pnpm","pnpm dlx skills add ozzycodes2\u002Fclaudellum --skill plugin-curation","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fozzycodes2\u002Fclaudellum\u002Fblob\u002Fmain\u002Fsession-improv\u002Fskills\u002Fplugin-curation\u002FSKILL.md",[46],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"queries.sql","OTHER","其他",2037,"https:\u002F\u002Fgithub.com\u002Fozzycodes2\u002Fclaudellum\u002Fblob\u002Fmain\u002Fsession-improv\u002Fskills\u002Fplugin-curation\u002Fqueries.sql",[53,60,67,74,81,88],{"skillNo":54,"name":55,"nameZh":6,"description":56,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":57,"categories":58,"keywords":59,"publishedAt":19},"01M7ASQB3CTKMKDNAXD9DCF1YC","writing-design-docs","Use when the user wants to write a problem-statement design doc, says \"let's plan a feature properly\" \u002F \"write a design doc\" \u002F \"I'm thinking about how to do X\" for a non-trivial X, or before committing to a non-obvious technical direction","https:\u002F\u002Fgithub.com\u002Fozzycodes2\u002Fclaudellum\u002Fblob\u002Fmain\u002Ftechnical-writing\u002Fskills\u002Fwriting-design-docs\u002FSKILL.md",[],[],{"skillNo":61,"name":62,"nameZh":6,"description":63,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":64,"categories":65,"keywords":66,"publishedAt":19},"01M7ASQ6JD4QRV5ZKK2QME401Y","write-doc","Use when the user asks to write or draft a README, ADR, runbook, tutorial, how-to, explanation, or reference doc, or a caller passes \"type ADR\" or \"type README\". Picks the type, fills a template, adds diagrams. RFCs and design docs belong to writing-design-docs.","https:\u002F\u002Fgithub.com\u002Fozzycodes2\u002Fclaudellum\u002Fblob\u002Fmain\u002Ftechnical-writing\u002Fskills\u002Fwrite-doc\u002FSKILL.md",[],[],{"skillNo":68,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":71,"categories":72,"keywords":73,"publishedAt":19},"01M7ASQ1VWNVP1RE7M5D4G2QMG","review-doc","Use when the user asks to review, critique, score, grade, or \"how good is this doc\" for a Markdown doc or pasted text. Read-only. Scores it on the 20 Practical Prose dimensions and emits a findings table. Does not edit; fix-doc applies the findings next.","https:\u002F\u002Fgithub.com\u002Fozzycodes2\u002Fclaudellum\u002Fblob\u002Fmain\u002Ftechnical-writing\u002Fskills\u002Freview-doc\u002FSKILL.md",[],[],{"skillNo":75,"name":76,"nameZh":6,"description":77,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":78,"categories":79,"keywords":80,"publishedAt":19},"01M7ASPWST208JZNR7HF7J6KMV","fix-doc","Use when the user asks to fix, edit for prose quality, polish, tighten, humanize, or de-slop a Markdown doc or pasted text, or says \"make this read less AI\". Applies review-doc findings, generates flagged diagrams, lists substantive issues for the author, then runs the AI-tell pass last. Modifies the target.","https:\u002F\u002Fgithub.com\u002Fozzycodes2\u002Fclaudellum\u002Fblob\u002Fmain\u002Ftechnical-writing\u002Fskills\u002Ffix-doc\u002FSKILL.md",[],[],{"skillNo":82,"name":83,"nameZh":6,"description":84,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":85,"categories":86,"keywords":87,"publishedAt":19},"01M7ASPS6Y9ZGK9KCDHNPDR75A","create-slides","Use when the user asks for a slide deck, presentation, talk, or slides from a document, report, PR, notes, or topic. Builds one self-contained Machinify-branded HTML deck in a minimalist style, with headline-plus-visual slides, speaker notes, and every diagram drawn by architecture-design:create-diagram.","https:\u002F\u002Fgithub.com\u002Fozzycodes2\u002Fclaudellum\u002Fblob\u002Fmain\u002Ftechnical-writing\u002Fskills\u002Fcreate-slides\u002FSKILL.md",[],[],{"skillNo":89,"name":90,"nameZh":6,"description":91,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":92,"categories":93,"keywords":94,"publishedAt":19},"01M7ASPAKFY045KHR92Y4QFTWV","start-session","Open a Claude Code working session in a fresh, isolated git worktree, so the user can start a task with a single \u002Fstart-session \u003Cname>. Use at the START of a session when the user says \"start session\", \"start a worktree\", \"spin up a worktree for ABC-123\", \"begin work on \u003Cticket>\", or passes a name\u002Fticket to work under. Creates and ENTERS a worktree via the EnterWorktree tool (never `git worktree add`) under .claude\u002Fworktrees\u002F, named exactly after the argument, on a local branch `worktree-\u003Cargument>`. Reports what the worktree inherited (`.worktreeinclude` copies, symlinked node_modules) and whether its `.env` still points at the main checkout's stack. If the argument is a JIRA key (e.g. ABC-123, PROJ-1505), also loads that story with the Atlassian CLI (`acli`) when available and summarizes the implementation ask. Also renames the session to the argument (the same effect as `\u002Frename \u003CNAME>`) so it's easy to find in `\u002Fresume`. Then stops and waits — it sets up the workspace and context, it does not implement.","https:\u002F\u002Fgithub.com\u002Fozzycodes2\u002Fclaudellum\u002Fblob\u002Fmain\u002Fsession-mgmt\u002Fskills\u002Fstart-session\u002FSKILL.md",[],[]]