[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M753CZAE1X6RVG36Z6ASSFF7":3,"skill-same:01M753CZAE1X6RVG36Z6ASSFF7":60},{"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":25,"commitSha":26,"skillDirectory":27,"forksCount":28,"repositoryHtmlUrl":29,"repositoryHomepageUrl":6,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":31,"install":32,"resources":45},"01M753CZAE1X6RVG36Z6ASSFF7","extend-repo-skill",null,"Extends an existing repository-specific Agent Skill with new capabilities, deeper coverage, troubleshooting, scripts, and usability tests. Use when the user asks to expand, improve, deepen, or add coverage to an already implemented skill instead of creating a new skill from scratch. If the repository itself changed and the old skill may be stale, use refresh-repo-skill instead.","VectorSpaceLab\u002FAREX-Skill","VectorSpaceLab","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F180637102?v=4","01M7538C5WPSJGWF7KHR4EFZ9F","01M7538C6R2GFP53NPXTCZNBRT",351,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fcli\u002Fpackages\u002Fcoding-agent\u002Fsrc\u002Fdisco\u002Fskills\u002Fextend-repo-skill\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: extend-repo-skill\ndescription: \"Extends an existing repository-specific Agent Skill with new capabilities, deeper coverage, troubleshooting, scripts, and usability tests. Use when the user asks to expand, improve, deepen, or add coverage to an already implemented skill instead of creating a new skill from scratch. If the repository itself changed and the old skill may be stale, use refresh-repo-skill instead.\"\nmetadata:\n  disco-role: meta\n---\n\n# Extend Repo Skill\n\n## Purpose\n\nUse this skill when the user already has a generated or hand-written Agent Skill and wants to expand it without losing the useful behavior it already provides.\n\nTypical requests include:\n\n- Add support for a new repo feature, public API, CLI, workflow, model family, config format, or troubleshooting path that the user specifically wants covered.\n- Improve thin references, missing scripts, poor routing, incomplete usability tests, or unclear triggering.\n- Extend an existing skill for a larger repo instead of regenerating it from scratch.\n\nIf the user's main concern is that the repository code, docs, APIs, CLIs,\nconfigs, or dependencies changed and the old skill may be stale, use the sibling\n`refresh-repo-skill` skill instead.\n\nThe expected result is an edited runtime skill directory plus updated\nusability test cases and a verification\u002Fhuman-review package under the\nskill's review\u002Ftest artifact directory. Do not create a separate replacement\nskill unless the user explicitly asks for a fork.\n\n## Inputs\n\nGather or infer:\n\n- Existing skill directory containing `SKILL.md`.\n- Resolved DisCo agent directory, normally `~\u002F.disco\u002Fagent` or the active\n  `DISCO_CODING_AGENT_DIR`, when the existing skill may be a live managed copy.\n- Repository path used as the evidence source for the extension.\n- The user's requested new capability, failure mode, coverage gap, or quality improvement.\n- Python inspection environment and installed package name when live API verification is needed.\n- Existing review\u002Ftest artifact directory, if any.\n- Desired review\u002Ftest artifact output directory, if the user has a preference.\n\nIf the existing skill directory is missing or does not contain `SKILL.md`, stop and ask for the correct skill path.\n\n## Reference Map\n\nRead these references as the workflow reaches each stage:\n\n- [references\u002Fextension-planning.md](references\u002Fextension-planning.md): scope capture, current-skill audit, repository evidence gathering, coverage-gap planning, and extension plan format.\n- [references\u002Fediting-and-versioning.md](references\u002Fediting-and-versioning.md): editing rules, preserving existing behavior, frontmatter constraints, references\u002Fscripts updates, and usability-case expansion.\n- [references\u002Fverification-and-handoff.md](references\u002Fverification-and-handoff.md): automatic verification, human review package, regression-sensitive checks, and final handoff.\n- [..\u002Fverify-repo-skill\u002Fscripts\u002Fresolve_repo_license.mjs](..\u002Fverify-repo-skill\u002Fscripts\u002Fresolve_repo_license.mjs)\n  and [..\u002Fverify-repo-skill\u002Fscripts\u002Fapply_repo_license.mjs](..\u002Fverify-repo-skill\u002Fscripts\u002Fapply_repo_license.mjs):\n  resolve and recursively apply the source repository license for the exact\n  extension commit before verification.\n\nWhen useful, also read sibling workflow-skill references: `..\u002Fcreate-repo-skill\u002Freferences\u002F` for canonical skill IDs, repository evidence discovery, installed-package inspection, planning, and writing, and `..\u002Fverify-repo-skill\u002Freferences\u002F` for usability test-case format, verification review, and import\u002Findex routing guidance.\n\n## Required Workflow\n\n1. Resolve the existing skill directory, repository path, requested extension,\n   DisCo agent directory, and review\u002Ftest artifact directory. If the user does\n   not specify an artifact directory, default to\n   `\u003Crepository-path>\u002Fskills\u002Ftests\u002F\u003Cskill-id>\u002F`, with usability cases under\n   `test-cases\u002F` and review reports under `reports\u002F`. If the existing skill is\n   the live managed copy under\n   `\u003Cagent-dir>\u002Fskills\u002Frepositories\u002Frepo-skills\u002F\u003Cskill-id>\u002F`, copy only its runtime tree to a\n   working directory outside `\u003Cagent-dir>\u002Fskills\u002F`, preserving the\n   `\u003Cskill-id>` directory basename. Keep that working copy until the dedicated\n   importer succeeds; never edit the live managed copy in place.\n2. Read [references\u002Fextension-planning.md](references\u002Fextension-planning.md). Audit current root\u002Fsub-skill routing, bundled references, bundled scripts, evals, usability tests, and known gaps before editing.\n3. Gather targeted repository evidence for the requested extension. Use source code plus installed-package inspection for API facts; use docs, examples, tests, and configs for intent and workflows.\n4. Write a concise extension plan that maps each new or changed capability to exactly one skill location, reference, script, or usability case.\n5. Read [references\u002Fediting-and-versioning.md](references\u002Fediting-and-versioning.md). Edit the resolved source or external working copy while preserving useful current guidance, frontmatter IDs, and public structure. Preserve the canonical `repo_id` and `skill_id` unless the user explicitly requests an identity migration. Treat the current area-family assignments as the baseline. If the extension is only deeper coverage of already routed capabilities, retain those assignments. If it adds a distinct capability, removes a capability, changes the repository's scope, changes the taxonomy hash, or the user requests reclassification, create a new external area-family routing handoff with assignment-specific evidence and update the minimal v2 metadata to match it. Never hand-edit generated family pages or silently change routing in the skill prose.\n6. Resolve and apply the repository license after the extension source commit\n   is known and before verification:\n\n   ```bash\n   node ..\u002Fverify-repo-skill\u002Fscripts\u002Fresolve_repo_license.mjs \\\n     --repository \u003Cowner\u002Frepository> \\\n     --source-commit \u003C40-hex-source-commit> \\\n     --json > \u003Cartifact-root>\u002Freports\u002Flicense-resolution.json\n   node ..\u002Fverify-repo-skill\u002Fscripts\u002Fapply_repo_license.mjs \\\n     --skill-dir \u003Cexternal-runtime-skill-dir> \\\n     --license \u003Cvalue-from-report>\n   ```\n\n   Re-query on every extension and synchronize the one returned value over the\n   complete root\u002Fsub-skill tree. Record old value, new value, source commit,\n   status, and reason in the handoff. Preserve `NOASSERTION` as the accepted\n   source value and use `NO_LICENSE` only when the resolver is unavailable; this\n   fallback is a warning, not a legal conclusion, and does not by itself block\n   verification or import.\n7. Add or update usability test cases for the new capability and at least one regression-sensitive existing workflow under `test-cases\u002F` in the review\u002Ftest artifact directory.\n8. Read [references\u002Fverification-and-handoff.md](references\u002Fverification-and-handoff.md). Run automatic verification, create a human review package under `reports\u002F` in the review\u002Ftest artifact directory, and fix blocking issues.\n9. After verification passes, follow `verify-repo-skill`'s structured import\n   policy. Ask for import or overwrite approval unless the user already\n   authorized that exact action, then run\n   `verify-repo-skill\u002Fscripts\u002Fimport_repo_skill.mjs` with the verified external\n   runtime directory and `--routing-entry\n   \u003Crepo-path>\u002Fskills\u002Fdisco\u002Frouting_decision\u002Fclassification.json`. The external\n   handoff is mandatory for a normal classified import, even when the existing\n   area-family assignments are retained. Pass `--overwrite` only for the exact\n   approved existing managed skill. The importer installs the runtime tree and\n   rebuilds the sibling live `repo-skills-router` under one rollback-capable\n   global lock.\n   After success, DisCo Researcher can use the extended skill in a new session;\n   use `import-repo-skills-to-agent` only for an explicitly requested\n   cross-agent export.\n\n## Non-Negotiables\n\n- Do not discard and rewrite the whole skill just because the extension is easier to express from scratch.\n- Do not edit a live skill under `\u003Cagent-dir>\u002Fskills\u002Frepositories\u002Frepo-skills\u002F` directly.\n  Extend an external working copy, then let\n  `verify-repo-skill\u002Fscripts\u002Fimport_repo_skill.mjs` replace the approved live\n  target.\n- Do not install an extended repo skill through a manually assembled copy and\n  router-update sequence.\n- Do not rename root or sub-skill IDs unless the user explicitly asks or the current names are invalid.\n- Do not silently change a repository's `repo_id` or `skill_id`; identity\n  changes are separate migrations that must update the central repository\n  index and all router references together.\n- Do not silently change area-family assignments. Preserve them for a\n  same-scope extension, and use the verified routing handoff plus the locked\n  importer when the extension changes capability scope or requires\n  reclassification.\n- Do not remove existing references, scripts, routes, or usability cases unless they are wrong, stale, duplicated, or replaced by a better self-contained artifact.\n- Do not add claims about APIs, CLIs, configs, data formats, or runtime behavior without repository evidence or live inspection.\n- Do not leak local checkout paths, Python executable paths, virtualenv or conda names, `pip show` locations, or machine-specific details into public skill files.\n- Do not link runtime skill documentation to original repo docs, examples, notebooks, or scripts. Distill or adapt material into the skill directory.\n- Do not leave new or existing runtime guidance that tells future agents to run,\n  read, or adapt source repo scripts by path. Bundle the needed script,\n  wrapper, or distilled recipe inside the skill's own `scripts\u002F` or\n  `references\u002F` tree before linking it.\n- Do not treat `tests\u002F` as a skill directory. It is the review\u002Ftest artifact area.\n- Do not write `evals\u002F`, verification reports, human-review notes, publication checklists, prompt samples, benchmark notes, or other check-only artifacts inside the runtime skill directory. Put concrete usability cases under the review\u002Ftest artifact directory's `test-cases\u002F` subtree and reports under its `reports\u002F` subtree, defaulting to `\u003Crepository-path>\u002Fskills\u002Ftests\u002F\u003Cskill-id>\u002F`.\n\n## Output Summary\n\nBy the end, the user should have:\n\n- The source skill directory or external working copy updated without mutating a\n  live managed skill outside the import transaction.\n- New or revised references\u002Fscripts\u002Fsub-skills where the requested capability needs depth.\n- Updated usability test cases in the configured review\u002Ftest artifact directory's `test-cases\u002F` subtree.\n- An automatic verification report and human review package in the configured review\u002Ftest artifact directory's `reports\u002F` subtree.\n- An approved managed import that DisCo Researcher can use directly in a new\n  session, or a clear staged-only status when import was declined.\n- A routing handoff that records whether the existing assignments were\n  preserved or reclassified, including evidence for any changed assignment.\n- A final handoff that distinguishes changed public skill content, review\u002Ftest\n  artifacts, evidence used, import status, and remaining gaps.\n","# Extend Repo Skill\n\n## Purpose\n\nUse this skill when the user already has a generated or hand-written Agent Skill and wants to expand it without losing the useful behavior it already provides.\n\nTypical requests include:\n\n- Add support for a new repo feature, public API, CLI, workflow, model family, config format, or troubleshooting path that the user specifically wants covered.\n- Improve thin references, missing scripts, poor routing, incomplete usability tests, or unclear triggering.\n- Extend an existing skill for a larger repo instead of regenerating it from scratch.\n\nIf the user's main concern is that the repository code, docs, APIs, CLIs,\nconfigs, or dependencies changed and the old skill may be stale, use the sibling\n`refresh-repo-skill` skill instead.\n\nThe expected result is an edited runtime skill directory plus updated\nusability test cases and a verification\u002Fhuman-review package under the\nskill's review\u002Ftest artifact directory. Do not create a separate replacement\nskill unless the user explicitly asks for a fork.\n\n## Inputs\n\nGather or infer:\n\n- Existing skill directory containing `SKILL.md`.\n- Resolved DisCo agent directory, normally `~\u002F.disco\u002Fagent` or the active\n  `DISCO_CODING_AGENT_DIR`, when the existing skill may be a live managed copy.\n- Repository path used as the evidence source for the extension.\n- The user's requested new capability, failure mode, coverage gap, or quality improvement.\n- Python inspection environment and installed package name when live API verification is needed.\n- Existing review\u002Ftest artifact directory, if any.\n- Desired review\u002Ftest artifact output directory, if the user has a preference.\n\nIf the existing skill directory is missing or does not contain `SKILL.md`, stop and ask for the correct skill path.\n\n## Reference Map\n\nRead these references as the workflow reaches each stage:\n\n- [references\u002Fextension-planning.md](references\u002Fextension-planning.md): scope capture, current-skill audit, repository evidence gathering, coverage-gap planning, and extension plan format.\n- [references\u002Fediting-and-versioning.md](references\u002Fediting-and-versioning.md): editing rules, preserving existing behavior, frontmatter constraints, references\u002Fscripts updates, and usability-case expansion.\n- [references\u002Fverification-and-handoff.md](references\u002Fverification-and-handoff.md): automatic verification, human review package, regression-sensitive checks, and final handoff.\n- [..\u002Fverify-repo-skill\u002Fscripts\u002Fresolve_repo_license.mjs](..\u002Fverify-repo-skill\u002Fscripts\u002Fresolve_repo_license.mjs)\n  and [..\u002Fverify-repo-skill\u002Fscripts\u002Fapply_repo_license.mjs](..\u002Fverify-repo-skill\u002Fscripts\u002Fapply_repo_license.mjs):\n  resolve and recursively apply the source repository license for the exact\n  extension commit before verification.\n\nWhen useful, also read sibling workflow-skill references: `..\u002Fcreate-repo-skill\u002Freferences\u002F` for canonical skill IDs, repository evidence discovery, installed-package inspection, planning, and writing, and `..\u002Fverify-repo-skill\u002Freferences\u002F` for usability test-case format, verification review, and import\u002Findex routing guidance.\n\n## Required Workflow\n\n1. Resolve the existing skill directory, repository path, requested extension,\n   DisCo agent directory, and review\u002Ftest artifact directory. If the user does\n   not specify an artifact directory, default to\n   `\u003Crepository-path>\u002Fskills\u002Ftests\u002F\u003Cskill-id>\u002F`, with usability cases under\n   `test-cases\u002F` and review reports under `reports\u002F`. If the existing skill is\n   the live managed copy under\n   `\u003Cagent-dir>\u002Fskills\u002Frepositories\u002Frepo-skills\u002F\u003Cskill-id>\u002F`, copy only its runtime tree to a\n   working directory outside `\u003Cagent-dir>\u002Fskills\u002F`, preserving the\n   `\u003Cskill-id>` directory basename. Keep that working copy until the dedicated\n   importer succeeds; never edit the live managed copy in place.\n2. Read [references\u002Fextension-planning.md](references\u002Fextension-planning.md). Audit current root\u002Fsub-skill routing, bundled references, bundled scripts, evals, usability tests, and known gaps before editing.\n3. Gather targeted repository evidence for the requested extension. Use source code plus installed-package inspection for API facts; use docs, examples, tests, and configs for intent and workflows.\n4. Write a concise extension plan that maps each new or changed capability to exactly one skill location, reference, script, or usability case.\n5. Read [references\u002Fediting-and-versioning.md](references\u002Fediting-and-versioning.md). Edit the resolved source or external working copy while preserving useful current guidance, frontmatter IDs, and public structure. Preserve the canonical `repo_id` and `skill_id` unless the user explicitly requests an identity migration. Treat the current area-family assignments as the baseline. If the extension is only deeper coverage of already routed capabilities, retain those assignments. If it adds a distinct capability, removes a capability, changes the repository's scope, changes the taxonomy hash, or the user requests reclassification, create a new external area-family routing handoff with assignment-specific evidence and update the minimal v2 metadata to match it. Never hand-edit generated family pages or silently change routing in the skill prose.\n6. Resolve and apply the repository license after the extension source commit\n   is known and before verification:\n\n   ```bash\n   node ..\u002Fverify-repo-skill\u002Fscripts\u002Fresolve_repo_license.mjs \\\n     --repository \u003Cowner\u002Frepository> \\\n     --source-commit \u003C40-hex-source-commit> \\\n     --json > \u003Cartifact-root>\u002Freports\u002Flicense-resolution.json\n   node ..\u002Fverify-repo-skill\u002Fscripts\u002Fapply_repo_license.mjs \\\n     --skill-dir \u003Cexternal-runtime-skill-dir> \\\n     --license \u003Cvalue-from-report>\n   ```\n\n   Re-query on every extension and synchronize the one returned value over the\n   complete root\u002Fsub-skill tree. Record old value, new value, source commit,\n   status, and reason in the handoff. Preserve `NOASSERTION` as the accepted\n   source value and use `NO_LICENSE` only when the resolver is unavailable; this\n   fallback is a warning, not a legal conclusion, and does not by itself block\n   verification or import.\n7. Add or update usability test cases for the new capability and at least one regression-sensitive existing workflow under `test-cases\u002F` in the review\u002Ftest artifact directory.\n8. Read [references\u002Fverification-and-handoff.md](references\u002Fverification-and-handoff.md). Run automatic verification, create a human review package under `reports\u002F` in the review\u002Ftest artifact directory, and fix blocking issues.\n9. After verification passes, follow `verify-repo-skill`'s structured import\n   policy. Ask for import or overwrite approval unless the user already\n   authorized that exact action, then run\n   `verify-repo-skill\u002Fscripts\u002Fimport_repo_skill.mjs` with the verified external\n   runtime directory and `--routing-entry\n   \u003Crepo-path>\u002Fskills\u002Fdisco\u002Frouting_decision\u002Fclassification.json`. The external\n   handoff is mandatory for a normal classified import, even when the existing\n   area-family assignments are retained. Pass `--overwrite` only for the exact\n   approved existing managed skill. The importer installs the runtime tree and\n   rebuilds the sibling live `repo-skills-router` under one rollback-capable\n   global lock.\n   After success, DisCo Researcher can use the extended skill in a new session;\n   use `import-repo-skills-to-agent` only for an explicitly requested\n   cross-agent export.\n\n## Non-Negotiables\n\n- Do not discard and rewrite the whole skill just because the extension is easier to express from scratch.\n- Do not edit a live skill under `\u003Cagent-dir>\u002Fskills\u002Frepositories\u002Frepo-skills\u002F` directly.\n  Extend an external working copy, then let\n  `verify-repo-skill\u002Fscripts\u002Fimport_repo_skill.mjs` replace the approved live\n  target.\n- Do not install an extended repo skill through a manually assembled copy and\n  router-update sequence.\n- Do not rename root or sub-skill IDs unless the user explicitly asks or the current names are invalid.\n- Do not silently change a repository's `repo_id` or `skill_id`; identity\n  changes are separate migrations that must update the central repository\n  index and all router references together.\n- Do not silently change area-family assignments. Preserve them for a\n  same-scope extension, and use the verified routing handoff plus the locked\n  importer when the extension changes capability scope or requires\n  reclassification.\n- Do not remove existing references, scripts, routes, or usability cases unless they are wrong, stale, duplicated, or replaced by a better self-contained artifact.\n- Do not add claims about APIs, CLIs, configs, data formats, or runtime behavior without repository evidence or live inspection.\n- Do not leak local checkout paths, Python executable paths, virtualenv or conda names, `pip show` locations, or machine-specific details into public skill files.\n- Do not link runtime skill documentation to original repo docs, examples, notebooks, or scripts. Distill or adapt material into the skill directory.\n- Do not leave new or existing runtime guidance that tells future agents to run,\n  read, or adapt source repo scripts by path. Bundle the needed script,\n  wrapper, or distilled recipe inside the skill's own `scripts\u002F` or\n  `references\u002F` tree before linking it.\n- Do not treat `tests\u002F` as a skill directory. It is the review\u002Ftest artifact area.\n- Do not write `evals\u002F`, verification reports, human-review notes, publication checklists, prompt samples, benchmark notes, or other check-only artifacts inside the runtime skill directory. Put concrete usability cases under the review\u002Ftest artifact directory's `test-cases\u002F` subtree and reports under its `reports\u002F` subtree, defaulting to `\u003Crepository-path>\u002Fskills\u002Ftests\u002F\u003Cskill-id>\u002F`.\n\n## Output Summary\n\nBy the end, the user should have:\n\n- The source skill directory or external working copy updated without mutating a\n  live managed skill outside the import transaction.\n- New or revised references\u002Fscripts\u002Fsub-skills where the requested capability needs depth.\n- Updated usability test cases in the configured review\u002Ftest artifact directory's `test-cases\u002F` subtree.\n- An automatic verification report and human review package in the configured review\u002Ftest artifact directory's `reports\u002F` subtree.\n- An approved managed import that DisCo Researcher can use directly in a new\n  session, or a clear staged-only status when import was declined.\n- A routing handoff that records whether the existing assignments were\n  preserved or reclassified, including evidence for any changed assignment.\n- A final handoff that distinguishes changed public skill content, review\u002Ftest\n  artifacts, evidence used, import status, and remaining gaps.\n",{"name":5,"metadata":23,"description":7},{"disco-role":24},"meta","main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","cli\u002Fpackages\u002Fcoding-agent\u002Fsrc\u002Fdisco\u002Fskills\u002Fextend-repo-skill",27,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab","2026-09-03 10:25:21",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add VectorSpaceLab\u002FAREX-Skill --skill extend-repo-skill",{"tool":38,"command":39},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill extend-repo-skill",{"tool":41,"command":42},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill extend-repo-skill","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fcli\u002Fpackages\u002Fcoding-agent\u002Fsrc\u002Fdisco\u002Fskills\u002Fextend-repo-skill\u002FSKILL.md",true,[46,52,56],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"references\u002Fediting-and-versioning.md","REFERENCE","参考文档",4941,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fcli\u002Fpackages\u002Fcoding-agent\u002Fsrc\u002Fdisco\u002Fskills\u002Fextend-repo-skill\u002Freferences\u002Fediting-and-versioning.md",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"references\u002Fextension-planning.md",5514,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fcli\u002Fpackages\u002Fcoding-agent\u002Fsrc\u002Fdisco\u002Fskills\u002Fextend-repo-skill\u002Freferences\u002Fextension-planning.md",{"path":57,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":58,"sourceHtmlUrl":59},"references\u002Fverification-and-handoff.md",7684,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fcli\u002Fpackages\u002Fcoding-agent\u002Fsrc\u002Fdisco\u002Fskills\u002Fextend-repo-skill\u002Freferences\u002Fverification-and-handoff.md",[61,69,76,83,90,97],{"skillNo":62,"name":63,"nameZh":6,"description":64,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":65,"categories":66,"keywords":67,"publishedAt":68},"01M799NYSH1R3RVAHSM2D88Y84","passnet-triton-opt","Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block\u002Fgrid\u002Fwarp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-triton-opt\u002FSKILL.md",[],[],"2026-11-15 14:19:17",{"skillNo":70,"name":71,"nameZh":6,"description":72,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":73,"categories":74,"keywords":75,"publishedAt":68},"01M799NS5APEC0N1FZ7XZVW6BN","passnet-solve","END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-solve\u002FSKILL.md",[],[],{"skillNo":77,"name":78,"nameZh":6,"description":79,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":82,"publishedAt":68},"01M799NNY9BKSM21V5R7QVTBJX","passnet-skill","PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under .\u002Fpass_dir\u002F, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-skill\u002FSKILL.md",[],[],{"skillNo":84,"name":85,"nameZh":6,"description":86,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":87,"categories":88,"keywords":89,"publishedAt":68},"01M799NHH6NP7Z1RK77KGXAN6S","passnet-pattern-fusion","Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()\u002Freplacement_args()\u002Freplacement_func() files, when a pass \"failed to match\", or when deciding how to fuse multiple kernels\u002Fops.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-pattern-fusion\u002FSKILL.md",[],[],{"skillNo":91,"name":92,"nameZh":6,"description":93,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":94,"categories":95,"keywords":96,"publishedAt":68},"01M799ND2J4WNC2PGAZ0A06NDF","passnet-orchestrate","Round\u002Fbatch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep\u002Frevert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-orchestrate\u002FSKILL.md",[],[],{"skillNo":98,"name":99,"nameZh":6,"description":100,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":101,"categories":102,"keywords":103,"publishedAt":68},"01M799N8SBTMY02ZDG2AZF5ZWE","passnet-feedback","Fast, deterministic feedback for PassNet work: pre-flight pattern\u002Fmatch verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-feedback\u002FSKILL.md",[],[]]