[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M72DZA2X3B9TE15WJMAT9348":3,"skill-same:01M72DZA2X3B9TE15WJMAT9348":124},{"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":33,"commitSha":34,"skillDirectory":35,"forksCount":14,"repositoryHtmlUrl":36,"repositoryHomepageUrl":37,"ownerHtmlUrl":38,"ownerBlogUrl":39,"ownerDescription":40,"updatedAt":41,"install":42,"resources":55},"01M72DZA2X3B9TE15WJMAT9348","rlm",null,"Use when a task involves long context, many files, PDFs, logs, transcripts, URLs, corpora, or anything requiring recursive decomposition over external context that won't fit comfortably in a single prompt. NOT for ordinary small edits, short Q&A, or single-file inspection.","wbratz\u002Fbillys-ai-skills","wbratz","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F6062728?v=4","01M72DZ0C1G73BCCR3TQ71PYZR","William Bratz","01M72DZ0CQJCZNJ273HCJ609TF",0,"MIT","https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002FSKILL.md",[],[],"2026-11-10 18:39:20",false,"---\nname: rlm\ndescription: Use when a task involves long context, many files, PDFs, logs, transcripts, URLs, corpora, or anything requiring recursive decomposition over external context that won't fit comfortably in a single prompt. NOT for ordinary small edits, short Q&A, or single-file inspection.\ntriggers:\n  - \"use RLM\"\n  - \"run RLM\"\n  - \"long context analysis\"\n  - \"analyze this corpus\"\n  - \"analyze these PDFs\"\n  - \"summarize this directory\"\n  - \"recursive language model\"\n  - \"\u002Frlm\"\n---\n\n# RLM Skill\n\nYou are operating an RLM (Recursive Language Model) workflow. RLM is **not** a bigger prompt - it is a loop where a controller LLM writes Python in a REPL, executes it against external context, fires sub-LLM calls, and emits a final answer.\n\n## When to invoke\n\n**Use RLM when:**\n- Context exceeds ~50KB or spans many files\n- Task requires aggregating across multiple sources\n- PDFs, logs, transcripts, or websites are involved\n- The user explicitly asks for it (\"use RLM\", \"\u002Frlm\")\n\n**Do not use RLM when:**\n- Task is a small local edit\n- Question can be answered by reading 1-2 files\n- User wants a quick code search (use Grep\u002FGlob\u002FRead)\n\n## Workflow (mandatory order)\n\n### 1. Detect the target\n\nParse the user's request to identify:\n- Target type: file \u002F directory \u002F PDF \u002F URL \u002F log \u002F glob\n- Target size: bytes, file count, estimated tokens\n- Question \u002F task\n\nIf ambiguous, ask one clarifying question before planning.\n\nSee `references\u002Ftarget-detection.md` for type-detection heuristics.\n\n### 2. Generate the plan (do not execute yet)\n\nRun the planner:\n\n```bash\npython ${CLAUDE_PLUGIN_ROOT}\u002Fskills\u002Frlm\u002Fscripts\u002Frlm_plan.py \\\n  --target \u003Cpath-or-url> \\\n  --prompt \"\u003Cthe user's question>\" \\\n  --mode auto\n```\n\nThe planner outputs a structured plan. Display it to the user verbatim:\n\n```\nTarget: \u003Cdescription>  (\u003Csize summary>)\nType: \u003Cdetected type>\nRecommended mode: \u003Cmin|default|max>\n  depth=N iterations=N timeout=Ns errors=N concurrency=N\nModel routing:\n  Root (depth 0):   \u003Cmodel>\n  Depth 1:          \u003Cmodel | ->\n  Depth 2:          \u003Cmodel | ->\n  llm_query:        claude-haiku-4-5-20251001\nPredicted fanout: ~N Haiku calls in M rounds at K-wide\nEst. cost: $X.XX - $Y.YY\nTrajectory log:   .rlm\u002Flogs\u002Frlm-\u003Cid>.jsonl\n\nApprove? (yes \u002F mode=\u003Cmin|default|max> \u002F cancel)\n```\n\n### 3. Wait for verbal approval\n\nDo not execute until the user says \"yes\", \"go\", \"approved\", or similar. Accept edits like `mode=max --max-budget=5` and re-display the updated plan.\n\n### 4. Execute\n\nProbe capabilities:\n\n```bash\npython ${CLAUDE_PLUGIN_ROOT}\u002Fskills\u002Frlm\u002Fscripts\u002Frlm_health.py\n```\n\nThen run the appropriate runner:\n\n- If `python_version >= 3.11` and `rlms` is importable: run `rlm_run.py` (canonical)\n- Otherwise: run `rlm_native.py` (Claude-native fallback) and tell the user that recursion is prompt-level only in this mode\n\n```bash\npython ${CLAUDE_PLUGIN_ROOT}\u002Fskills\u002Frlm\u002Fscripts\u002Frlm_run.py \\\n  --target \u003Cpath> \\\n  --prompt \"\u003Cquestion>\" \\\n  --mode \u003Cchosen-mode> \\\n  --log-dir .rlm\u002Flogs\n```\n\n### 5. Report\n\nParse the JSON answer and summarize for the user. Always include:\n- The final answer\n- The trajectory log path (so they can `\u002Frlm:inspect` it)\n- Usage summary (tokens, calls, wall-clock time)\n\nIf the answer looks weak or the `ok` field is false, suggest `\u002Frlm:inspect \u003Ctrajectory-id>` and consider re-running with `mode=max` after re-approval.\n\n## Critical rules (from RLM paper + community)\n\n- **Never put the raw context into the chat.** RLM's whole point is keeping context outside the model prompt. Pass paths, not contents.\n- **Prefer `llm_query_batched` over `rlm_query`** in the controller's REPL code. Most subtasks are extraction\u002Fclassification\u002Fsummary, which fan out cheap to Haiku.\n- **Prefer `FINAL_VAR(name)` over `FINAL(...)`** - bare `FINAL` parsing is brittle. The controller should create a named variable first, then emit `FINAL_VAR(name)`.\n- **Default sandbox is IPython subprocess.** It gives hard cell timeouts and kernel isolation. Use `local` only when the user explicitly opts in and the content is trusted.\n- **Redact obvious secrets** (API keys, tokens) from context before chunks are sent to submodels. The planner runs a regex sweep; warn if hits are detected.\n- **API keys never go into the REPL.** The `LMHandler` brokers calls host-side. If the user asks you to put `ANTHROPIC_API_KEY` in chunked context, refuse.\n\n## References\n\n- `references\u002Frlm-runtime-contract.md` - REPL contract, FINAL_VAR rules, message-history shape\n- `references\u002Fuse-cases.md` - Catalog of supported input types with recommended config\n- `references\u002Fmodel-routing.md` - Why Opus\u002FSonnet\u002FHaiku map to specific depths\n- `references\u002Ftarget-detection.md` - How to classify a target before planning\n- `references\u002Fmodes.md` - Min\u002Fdefault\u002Fmax profile reference\n","# RLM Skill\n\nYou are operating an RLM (Recursive Language Model) workflow. RLM is **not** a bigger prompt - it is a loop where a controller LLM writes Python in a REPL, executes it against external context, fires sub-LLM calls, and emits a final answer.\n\n## When to invoke\n\n**Use RLM when:**\n- Context exceeds ~50KB or spans many files\n- Task requires aggregating across multiple sources\n- PDFs, logs, transcripts, or websites are involved\n- The user explicitly asks for it (\"use RLM\", \"\u002Frlm\")\n\n**Do not use RLM when:**\n- Task is a small local edit\n- Question can be answered by reading 1-2 files\n- User wants a quick code search (use Grep\u002FGlob\u002FRead)\n\n## Workflow (mandatory order)\n\n### 1. Detect the target\n\nParse the user's request to identify:\n- Target type: file \u002F directory \u002F PDF \u002F URL \u002F log \u002F glob\n- Target size: bytes, file count, estimated tokens\n- Question \u002F task\n\nIf ambiguous, ask one clarifying question before planning.\n\nSee `references\u002Ftarget-detection.md` for type-detection heuristics.\n\n### 2. Generate the plan (do not execute yet)\n\nRun the planner:\n\n```bash\npython ${CLAUDE_PLUGIN_ROOT}\u002Fskills\u002Frlm\u002Fscripts\u002Frlm_plan.py \\\n  --target \u003Cpath-or-url> \\\n  --prompt \"\u003Cthe user's question>\" \\\n  --mode auto\n```\n\nThe planner outputs a structured plan. Display it to the user verbatim:\n\n```\nTarget: \u003Cdescription>  (\u003Csize summary>)\nType: \u003Cdetected type>\nRecommended mode: \u003Cmin|default|max>\n  depth=N iterations=N timeout=Ns errors=N concurrency=N\nModel routing:\n  Root (depth 0):   \u003Cmodel>\n  Depth 1:          \u003Cmodel | ->\n  Depth 2:          \u003Cmodel | ->\n  llm_query:        claude-haiku-4-5-20251001\nPredicted fanout: ~N Haiku calls in M rounds at K-wide\nEst. cost: $X.XX - $Y.YY\nTrajectory log:   .rlm\u002Flogs\u002Frlm-\u003Cid>.jsonl\n\nApprove? (yes \u002F mode=\u003Cmin|default|max> \u002F cancel)\n```\n\n### 3. Wait for verbal approval\n\nDo not execute until the user says \"yes\", \"go\", \"approved\", or similar. Accept edits like `mode=max --max-budget=5` and re-display the updated plan.\n\n### 4. Execute\n\nProbe capabilities:\n\n```bash\npython ${CLAUDE_PLUGIN_ROOT}\u002Fskills\u002Frlm\u002Fscripts\u002Frlm_health.py\n```\n\nThen run the appropriate runner:\n\n- If `python_version >= 3.11` and `rlms` is importable: run `rlm_run.py` (canonical)\n- Otherwise: run `rlm_native.py` (Claude-native fallback) and tell the user that recursion is prompt-level only in this mode\n\n```bash\npython ${CLAUDE_PLUGIN_ROOT}\u002Fskills\u002Frlm\u002Fscripts\u002Frlm_run.py \\\n  --target \u003Cpath> \\\n  --prompt \"\u003Cquestion>\" \\\n  --mode \u003Cchosen-mode> \\\n  --log-dir .rlm\u002Flogs\n```\n\n### 5. Report\n\nParse the JSON answer and summarize for the user. Always include:\n- The final answer\n- The trajectory log path (so they can `\u002Frlm:inspect` it)\n- Usage summary (tokens, calls, wall-clock time)\n\nIf the answer looks weak or the `ok` field is false, suggest `\u002Frlm:inspect \u003Ctrajectory-id>` and consider re-running with `mode=max` after re-approval.\n\n## Critical rules (from RLM paper + community)\n\n- **Never put the raw context into the chat.** RLM's whole point is keeping context outside the model prompt. Pass paths, not contents.\n- **Prefer `llm_query_batched` over `rlm_query`** in the controller's REPL code. Most subtasks are extraction\u002Fclassification\u002Fsummary, which fan out cheap to Haiku.\n- **Prefer `FINAL_VAR(name)` over `FINAL(...)`** - bare `FINAL` parsing is brittle. The controller should create a named variable first, then emit `FINAL_VAR(name)`.\n- **Default sandbox is IPython subprocess.** It gives hard cell timeouts and kernel isolation. Use `local` only when the user explicitly opts in and the content is trusted.\n- **Redact obvious secrets** (API keys, tokens) from context before chunks are sent to submodels. The planner runs a regex sweep; warn if hits are detected.\n- **API keys never go into the REPL.** The `LMHandler` brokers calls host-side. If the user asks you to put `ANTHROPIC_API_KEY` in chunked context, refuse.\n\n## References\n\n- `references\u002Frlm-runtime-contract.md` - REPL contract, FINAL_VAR rules, message-history shape\n- `references\u002Fuse-cases.md` - Catalog of supported input types with recommended config\n- `references\u002Fmodel-routing.md` - Why Opus\u002FSonnet\u002FHaiku map to specific depths\n- `references\u002Ftarget-detection.md` - How to classify a target before planning\n- `references\u002Fmodes.md` - Min\u002Fdefault\u002Fmax profile reference\n",{"name":5,"triggers":24,"description":7},[25,26,27,28,29,30,31,32],"use RLM","run RLM","long context analysis","analyze this corpus","analyze these PDFs","summarize this directory","recursive language model","\u002Frlm","main","94180703a15a9ead6d04c29c492368d6f5aa02db","claude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm","https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills","https:\u002F\u002Fwww.williambratz.com\u002Fprojects","https:\u002F\u002Fgithub.com\u002Fwbratz","https:\u002F\u002Fwww.williambratz.com","Senior software engineer building production AI platforms, MCP infrastructure, and distributed systems.","2026-11-01 12:09:23",{"source":8,"skillName":5,"commands":43,"prompt":53,"zipAvailable":54},[44,47,50],{"tool":45,"command":46},"npx","npx skills add wbratz\u002Fbillys-ai-skills --skill rlm",{"tool":48,"command":49},"bunx","bunx skills add wbratz\u002Fbillys-ai-skills --skill rlm",{"tool":51,"command":52},"pnpm","pnpm dlx skills add wbratz\u002Fbillys-ai-skills --skill rlm","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002FSKILL.md",true,[56,62,66,70,74,78,84,88,92,96,100,104,108,112,116,120],{"path":57,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":60,"sourceHtmlUrl":61},"references\u002Fmodel-routing.md","REFERENCE","参考文档",3786,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Freferences\u002Fmodel-routing.md",{"path":63,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":64,"sourceHtmlUrl":65},"references\u002Fmodes.md",3320,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Freferences\u002Fmodes.md",{"path":67,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":68,"sourceHtmlUrl":69},"references\u002Frlm-runtime-contract.md",3889,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Freferences\u002Frlm-runtime-contract.md",{"path":71,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":72,"sourceHtmlUrl":73},"references\u002Ftarget-detection.md",2155,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Freferences\u002Ftarget-detection.md",{"path":75,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":76,"sourceHtmlUrl":77},"references\u002Fuse-cases.md",4646,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Freferences\u002Fuse-cases.md",{"path":79,"resourceType":80,"resourceTypeDesc":81,"sizeBytes":82,"sourceHtmlUrl":83},"scripts\u002Floaders\u002F__init__.py","SCRIPT","脚本",346,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Fscripts\u002Floaders\u002F__init__.py",{"path":85,"resourceType":80,"resourceTypeDesc":81,"sizeBytes":86,"sourceHtmlUrl":87},"scripts\u002Floaders\u002Fload_dir.py",1998,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Fscripts\u002Floaders\u002Fload_dir.py",{"path":89,"resourceType":80,"resourceTypeDesc":81,"sizeBytes":90,"sourceHtmlUrl":91},"scripts\u002Floaders\u002Fload_file.py",1959,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Fscripts\u002Floaders\u002Fload_file.py",{"path":93,"resourceType":80,"resourceTypeDesc":81,"sizeBytes":94,"sourceHtmlUrl":95},"scripts\u002Floaders\u002Fload_logs.py",2475,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Fscripts\u002Floaders\u002Fload_logs.py",{"path":97,"resourceType":80,"resourceTypeDesc":81,"sizeBytes":98,"sourceHtmlUrl":99},"scripts\u002Floaders\u002Fload_pdf.py",2560,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Fscripts\u002Floaders\u002Fload_pdf.py",{"path":101,"resourceType":80,"resourceTypeDesc":81,"sizeBytes":102,"sourceHtmlUrl":103},"scripts\u002Floaders\u002Fload_url.py",3225,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Fscripts\u002Floaders\u002Fload_url.py",{"path":105,"resourceType":80,"resourceTypeDesc":81,"sizeBytes":106,"sourceHtmlUrl":107},"scripts\u002Frlm_health.py",2495,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Fscripts\u002Frlm_health.py",{"path":109,"resourceType":80,"resourceTypeDesc":81,"sizeBytes":110,"sourceHtmlUrl":111},"scripts\u002Frlm_native.py",5461,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Fscripts\u002Frlm_native.py",{"path":113,"resourceType":80,"resourceTypeDesc":81,"sizeBytes":114,"sourceHtmlUrl":115},"scripts\u002Frlm_plan.py",12662,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Fscripts\u002Frlm_plan.py",{"path":117,"resourceType":80,"resourceTypeDesc":81,"sizeBytes":118,"sourceHtmlUrl":119},"scripts\u002Frlm_run.py",8095,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Fscripts\u002Frlm_run.py",{"path":121,"resourceType":80,"resourceTypeDesc":81,"sizeBytes":122,"sourceHtmlUrl":123},"scripts\u002Fsetup.py",6674,"https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm\u002Fskills\u002Frlm\u002Fscripts\u002Fsetup.py",[125,132,139,145,152,159],{"skillNo":126,"name":127,"nameZh":6,"description":128,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":129,"categories":130,"keywords":131,"publishedAt":19},"01M72DZMY6D9D12FE3CAPVEFRJ","todo-skill-name","TODO describe exactly when Codex should use this skill and when it should not.","https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fopenai\u002Ftemplates\u002Fskill\u002FSKILL.md",[],[],{"skillNo":133,"name":134,"nameZh":6,"description":135,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":136,"categories":137,"keywords":138,"publishedAt":19},"01M72DZK2J8D6JYBWMP4N1E3H4","example-skill","TODO replace this with the plugin skill trigger.","https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fopenai\u002Ftemplates\u002Fplugin\u002Fskills\u002Fexample-skill\u002FSKILL.md",[],[],{"skillNo":140,"name":5,"nameZh":6,"description":141,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":142,"categories":143,"keywords":144,"publishedAt":19},"01M72DZH2S7V8AW6CHN15V18BF","Use when the user explicitly says \"use RLM\" or a task needs long-context recursive decomposition across many files, logs, transcripts, corpora, PDFs, or documents. Do not use for narrow edits, small single-file analysis, ordinary repo search, or tasks with no external context.","https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fopenai\u002Fplugins\u002Frlm-codex\u002Fskills\u002Frlm\u002FSKILL.md",[],[],{"skillNo":146,"name":147,"nameZh":6,"description":148,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":149,"categories":150,"keywords":151,"publishedAt":19},"01M72DZEDMFV84RHNP7J7DAHN7","openai-marketplace-maintainer","Use when adding, reviewing, or fixing Codex skills, plugins, custom subagents, marketplace entries, or registry metadata under this OpenAI marketplace folder.","https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fopenai\u002F.agents\u002Fskills\u002Fopenai-marketplace-maintainer\u002FSKILL.md",[],[],{"skillNo":153,"name":154,"nameZh":6,"description":155,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":156,"categories":157,"keywords":158,"publishedAt":19},"01M72DZD2VA2ZX1NEB8VT4SDDK","My Skill Name","One sentence describing what this skill does and when to invoke it.","https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Ftemplates\u002Fskill\u002Fskill.md",[],[],{"skillNo":160,"name":161,"nameZh":6,"description":162,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":163,"categories":164,"keywords":165,"publishedAt":19},"01M72DZ77P3PZEBBZE5418WA9A","rlm-eval","Local-only evaluator for rlm-auto routing decisions. Invoke on \u002Frlm-auto:status, when the user asks about router accuracy or false positives, or when they want the dogfood report. Never sends data off the machine.","https:\u002F\u002Fgithub.com\u002Fwbratz\u002Fbillys-ai-skills\u002Fblob\u002Fmain\u002Fclaude\u002Fplugins\u002Fbillys-rlm-auto\u002Fskills\u002Frlm-eval\u002FSKILL.md",[],[]]