[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M753D3TEQF0SD9MMBH8QDMQ9":3,"skill-same:01M753D3TEQF0SD9MMBH8QDMQ9":56},{"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},"01M753D3TEQF0SD9MMBH8QDMQ9","import-repo-skills-to-agent",null,"Export all or selected DisCo-managed repository skills, their regenerated repository-index.jsonl, and a scoped area\u002Ffamily repo-skills-router into Codex, Claude Code, or another agent skills directory. Use when the user asks to copy, install, synchronize, or merge repository skills into another agent or project. Resolve exact skill IDs, obtain explicit approval for replacements, then use the bundled transactional helper for validation, Codex policy injection, deterministic router merging, rollback, and resume.","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\u002Fimport-repo-skills-to-agent\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: import-repo-skills-to-agent\ndescription: \"Export all or selected DisCo-managed repository skills, their regenerated repository-index.jsonl, and a scoped area\u002Ffamily repo-skills-router into Codex, Claude Code, or another agent skills directory. Use when the user asks to copy, install, synchronize, or merge repository skills into another agent or project. Resolve exact skill IDs, obtain explicit approval for replacements, then use the bundled transactional helper for validation, Codex policy injection, deterministic router merging, rollback, and resume.\"\nmetadata:\n  disco-role: meta\n---\n\n# Import Repo Skills To Agent\n\nThis is a Creator-mode meta workflow. Export DisCo's managed repository-skill\ncollection to another agent only when the user explicitly requests a\ncross-agent export. Cross-agent export is an artifact-management operation,\nnot downstream research: the selected repository skills normally carry the\n`operating` role, but Creator may inspect them as export inputs without using\nthem as operating context or asking the user to switch to Researcher. DisCo\nalready uses its live collection through `repo-skills-router` and explicit\n`\u002Fskill:\u003Cname>` invocation; it does not need this workflow for its own runtime.\n\nUse the bundled helper for every filesystem mutation:\n\n```text\nscripts\u002Fexport_repo_skills_to_agent.mjs\n```\n\nDo not manually copy skills, concatenate indexes, or merge generated router\nMarkdown.\n\n## Collection layout\n\nTreat the source collection root as the directory containing these siblings:\n\n```text\n\u003Cdisco-agent-root>\u002Fskills\u002Frepositories\u002F\n├── repo-skills\u002F\n│   ├── \u003Cskill-id>\u002F\n│   └── repository-index.jsonl\n└── repo-skills-router\u002F\n```\n\nThe default source agent root is `$DISCO_CODING_AGENT_DIR` when set, otherwise\n`~\u002F.disco\u002Fagent`. The target uses the same collection shape below its skills\nroot:\n\n```text\n\u003Ctarget-skills-root>\u002Frepositories\u002F\n├── repo-skills\u002F\n│   ├── \u003Cskill-id>\u002F\n│   └── repository-index.jsonl\n└── repo-skills-router\u002F\n```\n\nKeep unrelated target skills outside `repositories\u002F` unchanged.\n\n## Resolve the target\n\nPrefer an unambiguous target argument:\n\n- Use `--target-skills-root \u003Cdir>` when the path is already a skills root,\n  such as `~\u002F.agents\u002Fskills` or `~\u002F.claude\u002Fskills`.\n- Use `--target-agent-dir \u003Cdir>` when the path is an agent root; the helper\n  appends `skills\u002Frepositories`.\n- Use `--target \u003Cdir> --target-kind skills-root|agent-root` for a generic path.\n- Allow `--target` with automatic inference only when the basename or existing\n  layout makes the meaning unambiguous.\n\nFor Codex, default an agent-only request to the current user-level agent root\n`~\u002F.agents`, not the legacy `~\u002F.codex`. Pass `--target-agent codex`. For Claude\nCode or a generic target, pass `--target-agent agent-neutral` unless that\ntarget explicitly requires Codex policy metadata.\n\n## Resolve the selection\n\nThe helper accepts exact repository skill IDs. Before invoking it:\n\n1. Read the source `repo-skills\u002Frepository-index.jsonl` and direct\n   `repo-skills\u002F\u003Cskill-id>\u002FSKILL.md` files as export input artifacts. In\n   Creator, this is an explicit exception to the operating-context visibility\n   boundary: inspect them only to resolve and validate the export, never to\n   invoke their instructions or perform their downstream workflows.\n2. Resolve user terms, repository names, aliases, or wildcards to exact\n   `skill_id` values.\n3. Show the matched IDs when the request is ambiguous or a wildcard expands to\n   more than the user likely expects.\n4. Omit `--include-skill` for a full export. Repeat it, or pass comma-separated\n   IDs, for a subset.\n\nDo not select review artifacts, test outputs, environments, sessions, prompts,\nor directories outside the managed `repo-skills\u002F` collection.\n\n## Resolve conflicts before mutation\n\nInspect the target collection before starting the helper. For every selected\nskill ID already present under target `repositories\u002Frepo-skills\u002F`:\n\n1. Summarize the source and target paths and the relevant identity difference.\n2. Ask whether to replace that exact target skill.\n3. Pass `--overwrite-skill \u003Cskill-id>` only for approved replacements.\n4. Exclude a declined skill from the selected export and report it as skipped.\n\nNever silently overwrite a repository skill. The helper independently rejects\nan unapproved conflict and refuses repository-identity collisions.\n\n## Run the transaction\n\nFor a selected Codex export:\n\n```bash\nnode \u003Cthis-skill>\u002Fscripts\u002Fexport_repo_skills_to_agent.mjs \\\n  --source-agent-dir \u003Cdisco-agent-root> \\\n  --target-agent-dir ~\u002F.agents \\\n  --target-agent codex \\\n  --include-skill \u003Cskill-id> \\\n  --overwrite-skill \u003Capproved-skill-id>\n```\n\nFor a full agent-neutral export to an exact skills root:\n\n```bash\nnode \u003Cthis-skill>\u002Fscripts\u002Fexport_repo_skills_to_agent.mjs \\\n  --source-agent-dir \u003Cdisco-agent-root> \\\n  --target-skills-root \u003Ctarget-skills-root> \\\n  --target-agent agent-neutral\n```\n\nThe helper performs one deterministic workflow:\n\n1. Validate source and existing target paths, symlink safety, skill identity,\n   v2 `references\u002Frepo-routing-metadata.json`, exact taxonomy assignments, and\n   source\u002Ftarget non-overlap.\n2. Generate a filtered source router for the selected skills without modifying\n   the source collection.\n3. Merge selected source records with unrelated target records by canonical\n   `skill_id` and `owner\u002Frepository` identity.\n4. Stage the final `repo-skills\u002F` tree.\n5. For a Codex target, add or update `agents\u002Fopenai.yaml` beside every root and\n   descendant `SKILL.md` in the staged non-router repository skills, setting\n   `policy.allow_implicit_invocation: false` while preserving unrelated OpenAI\n   metadata. Never add this policy to `repo-skills-router`.\n6. Regenerate `repo-skills\u002Frepository-index.jsonl` and the sibling router from\n   the final merged records. Do not copy a stale source root index or combine\n   Markdown pages.\n7. Revalidate taxonomy paths, links, repository and assignment\n   indexes, router visibility, selected\u002Funselected boundaries, and Codex\n   policy before touching the live target.\n8. Replace only target `repositories\u002Frepo-skills\u002F` and\n   `repositories\u002Frepo-skills-router\u002F` through the persisted transaction.\n9. Validate the installed target and commit, or restore the exact previous\n   target if a post-mutation step fails.\n\nThe final root `repository-index.jsonl` must be byte-for-byte consistent with\n`repo-skills-router\u002Freferences\u002Findex\u002Frepositories.jsonl`. A subset export must\nnot leak unselected source skills into either index or any generated area or\nfamily page. Existing unrelated target repository skills and assignments must\nremain present.\n\n## Index and visibility contract\n\nThe long-lived repository indexes intentionally do not store a per-repository\nskill `content_sha256`. They preserve repository identity, provenance, routing\nassignments, and descriptions; their complete-file integrity is protected by\nthe router's `build-metadata.json` index digests. The import handoff may still\ncontain a one-time `skill_content_sha256`, which is checked while importing the\nverified runtime tree and is not copied into either long-lived repository index.\n\nTarget-specific `agents\u002Fopenai.yaml` files are still excluded from the\nhandoff's portable-tree digest. Do not ignore arbitrary extra files, symlinks,\nor non-regular entries.\n\nKeep `repo-skills-router` model-visible for a fresh target. Preserve an\nexisting target router's visibility by default; use\n`--router-visibility enabled|disabled` only when the user explicitly asks to\nchange it. Repository skills remain hidden from implicit selection through\ntheir portable frontmatter and, for Codex targets, the target-only OpenAI\npolicy.\n\n## Recover an interrupted export\n\nThe helper prints the persisted transaction directory on failure. Its\nmanifest stores both the selected index fingerprint and a transaction-only\nsnapshot of the selected source skill trees. This keeps interrupted exports\nsafe without reintroducing a per-repository digest into the long-lived\nindexes. Resume or retry exactly that transaction with:\n\n```bash\nnode \u003Cthis-skill>\u002Fscripts\u002Fexport_repo_skills_to_agent.mjs \\\n  --resume \u003Ctransaction-directory>\n```\n\nDo not manually repair a partial target. The helper restores any mutation\nphase to the recorded target snapshot, then reuses the persisted source,\nselection, overwrite approvals, target policy, and router visibility. For a\nvalidated pre-mutation transaction, it revalidates the source fingerprint,\ntarget snapshot, and staging before committing it.\nArguments do not need to be repeated; any explicitly repeated argument must\nmatch the manifest or resume stops before mutation.\n\nIf rollback cannot restore the exact recorded target snapshot, preserve the\nreported staging and backup paths and stop for manual inspection.\n\n## Safety and handoff\n\n- Keep the source collection read-only.\n- Reject source\u002Ftarget overlap and paths that traverse symlinks.\n- Do not import a structurally invalid selected skill or infer missing routing\n  metadata.\n- Do not delete unrelated target skills or assignments.\n- Do not add Codex policy to the router.\n- Do not report success until live target validation and transaction commit\n  both complete.\n\nReport the resolved source and target, exact selected IDs, copied and approved\nreplacement counts, final repository and assignment counts, router action,\nCodex policy mode, and any skipped conflicts. If the target agent caches skill\ndiscovery, tell the user to restart or reload it.\n","# Import Repo Skills To Agent\n\nThis is a Creator-mode meta workflow. Export DisCo's managed repository-skill\ncollection to another agent only when the user explicitly requests a\ncross-agent export. Cross-agent export is an artifact-management operation,\nnot downstream research: the selected repository skills normally carry the\n`operating` role, but Creator may inspect them as export inputs without using\nthem as operating context or asking the user to switch to Researcher. DisCo\nalready uses its live collection through `repo-skills-router` and explicit\n`\u002Fskill:\u003Cname>` invocation; it does not need this workflow for its own runtime.\n\nUse the bundled helper for every filesystem mutation:\n\n```text\nscripts\u002Fexport_repo_skills_to_agent.mjs\n```\n\nDo not manually copy skills, concatenate indexes, or merge generated router\nMarkdown.\n\n## Collection layout\n\nTreat the source collection root as the directory containing these siblings:\n\n```text\n\u003Cdisco-agent-root>\u002Fskills\u002Frepositories\u002F\n├── repo-skills\u002F\n│   ├── \u003Cskill-id>\u002F\n│   └── repository-index.jsonl\n└── repo-skills-router\u002F\n```\n\nThe default source agent root is `$DISCO_CODING_AGENT_DIR` when set, otherwise\n`~\u002F.disco\u002Fagent`. The target uses the same collection shape below its skills\nroot:\n\n```text\n\u003Ctarget-skills-root>\u002Frepositories\u002F\n├── repo-skills\u002F\n│   ├── \u003Cskill-id>\u002F\n│   └── repository-index.jsonl\n└── repo-skills-router\u002F\n```\n\nKeep unrelated target skills outside `repositories\u002F` unchanged.\n\n## Resolve the target\n\nPrefer an unambiguous target argument:\n\n- Use `--target-skills-root \u003Cdir>` when the path is already a skills root,\n  such as `~\u002F.agents\u002Fskills` or `~\u002F.claude\u002Fskills`.\n- Use `--target-agent-dir \u003Cdir>` when the path is an agent root; the helper\n  appends `skills\u002Frepositories`.\n- Use `--target \u003Cdir> --target-kind skills-root|agent-root` for a generic path.\n- Allow `--target` with automatic inference only when the basename or existing\n  layout makes the meaning unambiguous.\n\nFor Codex, default an agent-only request to the current user-level agent root\n`~\u002F.agents`, not the legacy `~\u002F.codex`. Pass `--target-agent codex`. For Claude\nCode or a generic target, pass `--target-agent agent-neutral` unless that\ntarget explicitly requires Codex policy metadata.\n\n## Resolve the selection\n\nThe helper accepts exact repository skill IDs. Before invoking it:\n\n1. Read the source `repo-skills\u002Frepository-index.jsonl` and direct\n   `repo-skills\u002F\u003Cskill-id>\u002FSKILL.md` files as export input artifacts. In\n   Creator, this is an explicit exception to the operating-context visibility\n   boundary: inspect them only to resolve and validate the export, never to\n   invoke their instructions or perform their downstream workflows.\n2. Resolve user terms, repository names, aliases, or wildcards to exact\n   `skill_id` values.\n3. Show the matched IDs when the request is ambiguous or a wildcard expands to\n   more than the user likely expects.\n4. Omit `--include-skill` for a full export. Repeat it, or pass comma-separated\n   IDs, for a subset.\n\nDo not select review artifacts, test outputs, environments, sessions, prompts,\nor directories outside the managed `repo-skills\u002F` collection.\n\n## Resolve conflicts before mutation\n\nInspect the target collection before starting the helper. For every selected\nskill ID already present under target `repositories\u002Frepo-skills\u002F`:\n\n1. Summarize the source and target paths and the relevant identity difference.\n2. Ask whether to replace that exact target skill.\n3. Pass `--overwrite-skill \u003Cskill-id>` only for approved replacements.\n4. Exclude a declined skill from the selected export and report it as skipped.\n\nNever silently overwrite a repository skill. The helper independently rejects\nan unapproved conflict and refuses repository-identity collisions.\n\n## Run the transaction\n\nFor a selected Codex export:\n\n```bash\nnode \u003Cthis-skill>\u002Fscripts\u002Fexport_repo_skills_to_agent.mjs \\\n  --source-agent-dir \u003Cdisco-agent-root> \\\n  --target-agent-dir ~\u002F.agents \\\n  --target-agent codex \\\n  --include-skill \u003Cskill-id> \\\n  --overwrite-skill \u003Capproved-skill-id>\n```\n\nFor a full agent-neutral export to an exact skills root:\n\n```bash\nnode \u003Cthis-skill>\u002Fscripts\u002Fexport_repo_skills_to_agent.mjs \\\n  --source-agent-dir \u003Cdisco-agent-root> \\\n  --target-skills-root \u003Ctarget-skills-root> \\\n  --target-agent agent-neutral\n```\n\nThe helper performs one deterministic workflow:\n\n1. Validate source and existing target paths, symlink safety, skill identity,\n   v2 `references\u002Frepo-routing-metadata.json`, exact taxonomy assignments, and\n   source\u002Ftarget non-overlap.\n2. Generate a filtered source router for the selected skills without modifying\n   the source collection.\n3. Merge selected source records with unrelated target records by canonical\n   `skill_id` and `owner\u002Frepository` identity.\n4. Stage the final `repo-skills\u002F` tree.\n5. For a Codex target, add or update `agents\u002Fopenai.yaml` beside every root and\n   descendant `SKILL.md` in the staged non-router repository skills, setting\n   `policy.allow_implicit_invocation: false` while preserving unrelated OpenAI\n   metadata. Never add this policy to `repo-skills-router`.\n6. Regenerate `repo-skills\u002Frepository-index.jsonl` and the sibling router from\n   the final merged records. Do not copy a stale source root index or combine\n   Markdown pages.\n7. Revalidate taxonomy paths, links, repository and assignment\n   indexes, router visibility, selected\u002Funselected boundaries, and Codex\n   policy before touching the live target.\n8. Replace only target `repositories\u002Frepo-skills\u002F` and\n   `repositories\u002Frepo-skills-router\u002F` through the persisted transaction.\n9. Validate the installed target and commit, or restore the exact previous\n   target if a post-mutation step fails.\n\nThe final root `repository-index.jsonl` must be byte-for-byte consistent with\n`repo-skills-router\u002Freferences\u002Findex\u002Frepositories.jsonl`. A subset export must\nnot leak unselected source skills into either index or any generated area or\nfamily page. Existing unrelated target repository skills and assignments must\nremain present.\n\n## Index and visibility contract\n\nThe long-lived repository indexes intentionally do not store a per-repository\nskill `content_sha256`. They preserve repository identity, provenance, routing\nassignments, and descriptions; their complete-file integrity is protected by\nthe router's `build-metadata.json` index digests. The import handoff may still\ncontain a one-time `skill_content_sha256`, which is checked while importing the\nverified runtime tree and is not copied into either long-lived repository index.\n\nTarget-specific `agents\u002Fopenai.yaml` files are still excluded from the\nhandoff's portable-tree digest. Do not ignore arbitrary extra files, symlinks,\nor non-regular entries.\n\nKeep `repo-skills-router` model-visible for a fresh target. Preserve an\nexisting target router's visibility by default; use\n`--router-visibility enabled|disabled` only when the user explicitly asks to\nchange it. Repository skills remain hidden from implicit selection through\ntheir portable frontmatter and, for Codex targets, the target-only OpenAI\npolicy.\n\n## Recover an interrupted export\n\nThe helper prints the persisted transaction directory on failure. Its\nmanifest stores both the selected index fingerprint and a transaction-only\nsnapshot of the selected source skill trees. This keeps interrupted exports\nsafe without reintroducing a per-repository digest into the long-lived\nindexes. Resume or retry exactly that transaction with:\n\n```bash\nnode \u003Cthis-skill>\u002Fscripts\u002Fexport_repo_skills_to_agent.mjs \\\n  --resume \u003Ctransaction-directory>\n```\n\nDo not manually repair a partial target. The helper restores any mutation\nphase to the recorded target snapshot, then reuses the persisted source,\nselection, overwrite approvals, target policy, and router visibility. For a\nvalidated pre-mutation transaction, it revalidates the source fingerprint,\ntarget snapshot, and staging before committing it.\nArguments do not need to be repeated; any explicitly repeated argument must\nmatch the manifest or resume stops before mutation.\n\nIf rollback cannot restore the exact recorded target snapshot, preserve the\nreported staging and backup paths and stop for manual inspection.\n\n## Safety and handoff\n\n- Keep the source collection read-only.\n- Reject source\u002Ftarget overlap and paths that traverse symlinks.\n- Do not import a structurally invalid selected skill or infer missing routing\n  metadata.\n- Do not delete unrelated target skills or assignments.\n- Do not add Codex policy to the router.\n- Do not report success until live target validation and transaction commit\n  both complete.\n\nReport the resolved source and target, exact selected IDs, copied and approved\nreplacement counts, final repository and assignment counts, router action,\nCodex policy mode, and any skipped conflicts. If the target agent caches skill\ndiscovery, tell the user to restart or reload it.\n",{"name":5,"metadata":23,"description":7},{"disco-role":24},"meta","main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","cli\u002Fpackages\u002Fcoding-agent\u002Fsrc\u002Fdisco\u002Fskills\u002Fimport-repo-skills-to-agent",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 import-repo-skills-to-agent",{"tool":38,"command":39},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill import-repo-skills-to-agent",{"tool":41,"command":42},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill import-repo-skills-to-agent","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fcli\u002Fpackages\u002Fcoding-agent\u002Fsrc\u002Fdisco\u002Fskills\u002Fimport-repo-skills-to-agent\u002FSKILL.md",true,[46,52],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"scripts\u002Fapply_codex_openai_policy.py","SCRIPT","脚本",3410,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fcli\u002Fpackages\u002Fcoding-agent\u002Fsrc\u002Fdisco\u002Fskills\u002Fimport-repo-skills-to-agent\u002Fscripts\u002Fapply_codex_openai_policy.py",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"scripts\u002Fexport_repo_skills_to_agent.mjs",58001,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fcli\u002Fpackages\u002Fcoding-agent\u002Fsrc\u002Fdisco\u002Fskills\u002Fimport-repo-skills-to-agent\u002Fscripts\u002Fexport_repo_skills_to_agent.mjs",[57,65,72,79,86,93],{"skillNo":58,"name":59,"nameZh":6,"description":60,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":61,"categories":62,"keywords":63,"publishedAt":64},"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":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":64},"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":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":64},"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":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":64},"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":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":64},"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":94,"name":95,"nameZh":6,"description":96,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":97,"categories":98,"keywords":99,"publishedAt":64},"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",[],[]]