[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M758BCRCST4D7P571FT30JNY":3,"skill-same:01M758BCRCST4D7P571FT30JNY":50},{"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":27,"commitSha":28,"skillDirectory":29,"forksCount":30,"repositoryHtmlUrl":31,"repositoryHomepageUrl":6,"ownerHtmlUrl":32,"ownerBlogUrl":33,"ownerDescription":34,"updatedAt":35,"install":36,"resources":49},"01M758BCRCST4D7P571FT30JNY","skillset-fix",null,"Diagnose and clear generated-output trouble in a Skillset repository — drift reported by check, a missing or hand-edited generated file, a conflicted path under a managed output root, or a halted rebase or merge. Routes to skillset resolve, skillset build --repair, and skillset check --only outputs. Use when a generated file or a skillset.lock differs from source, is missing, or is conflicted. Do not use for conflicts in authored source.","outfitter-dev\u002Fskillset","outfitter-dev","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F206289526?v=4","01M7589FBAPHBH5CX8XMH1MPJV","Outfitter","01M7589FCJ31KF5STHVN4XKY79",1,"MIT","https:\u002F\u002Fgithub.com\u002Foutfitter-dev\u002Fskillset\u002Fblob\u002Fmain\u002F.cursor\u002Fskills\u002Fskillset-fix\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\ndescription: Diagnose and clear generated-output trouble in a Skillset repository — drift reported by check, a missing or hand-edited generated file, a conflicted path under a managed output root, or a halted rebase or merge. Routes to skillset resolve, skillset build --repair, and skillset check --only outputs. Use when a generated file or a skillset.lock differs from source, is missing, or is conflicted. Do not use for conflicts in authored source.\nmetadata:\n  skillset.schema: \"1\"\n  version: 0.4.0\nname: skillset-fix\n---\n\n# Fix Skillset Generated Output\n\nSkillset commits its generated output. Those files are derived and\nnon-authoritative: their final content must come from authored source, never from\nmanual reconciliation. A conflict carries no valid merged content. A diff can\nstill preserve evidence of a misdirected human edit, so keep it until Skillset\nreports an `output-edited` or `diverged` verdict and the intent is recovered.\n\nNever leave an edit in a generated file, and never hand-merge one.\n\n## Run the repository's own compiler\n\nFirst check whether the repository builds Skillset itself by looking for\n`apps\u002Fskillset\u002Fpackage.json` with package name `skillset`.\n\n- In a Skillset compiler checkout, compare `skillset --version` with the version\n  in `apps\u002Fskillset\u002Fpackage.json`, then invoke the compiler through the repository:\n  `bun .\u002Fapps\u002Fskillset\u002Fsrc\u002Fcli.ts \u003Ccommand>`, or the `bun run skillset:*` scripts.\n- In a consumer repository, use the installed binary. MUST NOT compare it with the\n  root `package.json`; that version belongs to the consuming application.\n\nWhen the compiler-checkout versions differ, a globally installed `skillset` is\nshadowing the one under test, and its answers describe a different compiler.\n\nThis is not hypothetical. A stale 0.26.1 binary reporting `build is blocked` on a\nrepository whose own compiler restores the file correctly is what produced the\nfalse premise behind SET-599.\n\n## Decide whether the path is generated\n\nDo this first. It determines everything that follows.\n\n1. Run `skillset explain \u003Cpath> --root .`. It reports ownership and provenance, and\n   labels a derived path `(generated)`.\n2. When `explain` is unavailable, treat a path as generated if it appears as an\n   `outputPath` or in `files[]` of a `skillset.lock`, or is a `skillset.lock`.\n3. When the path is authored source, this skill does not apply. Resolve it as\n   ordinary source.\n\n## Read the verdict\n\n`skillset build --repair` reports repair verdicts and actions. A dry-run\n`skillset resolve` instead prints the conflict partition (`generated`, authored,\nor hand-edited); after `--yes` reaches the repair preview, it prints any\nnon-clean repair verdicts before its summary. Read the command's exact report\nrather than re-deriving it.\n\n| Verdict | What it means | What happens |\n| --- | --- | --- |\n| `clean` | The file already equals what source produces. | Nothing. |\n| `output-missing` | A managed file is absent. | Restored. No edit can be lost. |\n| `source-ahead` | Source moved; the output is stale. | Regenerated. |\n| `output-obsolete` | The lock claims it but source no longer produces it. | Removed. |\n| `output-edited` | The generated file was edited; source was not. | Preserved, not overwritten. See `## Recover a misdirected edit`. |\n| `lock-untrusted` | The lock records no verdict this repair can rely on. | Preserved. The fault is the lock, not the output. See `## Rebuild an untrusted lock`. |\n| `diverged` | The generated file **and** its source both moved. | Refused. See `## Escalate`. |\n\nThe verdicts come from a three-way comparison against the `outputHash` the lock\nrecorded for the last build — the file on disk, the lock, and what a fresh render\nwould produce.\n\nMUST NOT use file modification times to decide any of this. A rebase or a checkout\nrewrites mtimes across the whole tree, so in the situation where drift matters\nmost, every file appears equally fresh.\n\n`outputHash` is recorded per lock *item*, which is a group of files, so a single\nedited file marks its whole group. A path whose bytes already equal the render is\nreported `clean` regardless, because writing it would change nothing.\n\n## Clear a conflict during a rebase or merge\n\n1. Run `skillset resolve --root .` to see the partition without writing anything.\n2. When it names **authored** conflicts, it stages nothing and exits non-zero.\n   Resolve those files, `git add` them, then return to step 1. Regenerating from\n   source that still carries conflict markers would render those markers into\n   generated output.\n3. When it reports paths **edited by hand on one side of the conflict**, it\n   stages nothing and exits non-zero. Follow `## Recover a misdirected edit` for\n   each named path, run `git add \u003Csource-path>` for the recovered source change,\n   then return to step 1. MUST NOT commit while generated paths are still\n   unmerged; continuation creates the rebased or merged commit after generated\n   output is repaired.\n4. Run `skillset resolve --root . --yes`. It repairs the generated paths and stages\n   them.\n5. When it exits zero, continue the operation with the command that owns it:\n   `gh stack rebase --continue` in a stack, `git rebase --continue` for a rebase, or\n   `git merge --continue` for a merge.\n6. When it reports a `diverged` path, follow `## Escalate`.\n\nA conflicted file on disk represents neither side, so the verdict table has no\nbaseline there. `skillset resolve` instead checks each conflict side against the\nlock that side committed, and refuses when either disagrees. MUST NOT resolve\nsuch a path by taking a side. The pre-rebase commit still holds the edit; recover\nit from there.\n\n`skillset resolve` reads lock ownership from exact index blobs: stage 0 for an\nunconflicted lock and stages 2 and 3 for the two conflict sides. It never trusts\nmutable worktree lock bytes as provenance, so a lock carrying markers does not\nneed fixing by hand first. It stages only paths it classified as generated, and\nMUST NOT be used to stage authored source.\n\nWith `--yes`, resolve refuses worktree-only compiler inputs that could change the\nprojection: source, workspace config and manifest files, external repository\npackage sources, and explicit feature inputs. Staged inputs and unrelated dirt\noutside that input inventory do not block it.\n\nThe repair is transactional across the whole projection, not only the paths that\nhappened to conflict. Before writing, resolve records each affected path's exact\nbytes, executable mode, absence, missing parent directories, or symlink target.\nIf validation, build, or staging fails, it restores those preimages rather than\nleaving a partially repaired worktree.\n\n## Repair drift outside a conflict\n\nUse this when `skillset check` reports drift and no rebase or merge is in progress.\n\n1. Run `skillset build --repair --root .`. It classifies every managed path against\n   filesystem reality and prints a verdict for each that needs action.\n2. When every verdict is `clean`, `output-missing`, `source-ahead`, or\n   `output-obsolete`, run `skillset build --repair --root . --yes`. It restores,\n   regenerates, and removes.\n3. When any verdict is `output-edited`, it writes nothing. Follow\n   `## Recover a misdirected edit` for each named path, then return to step 1.\n4. When any verdict is `lock-untrusted`, it writes nothing. Follow\n   `## Rebuild an untrusted lock`, then return to step 1.\n5. When any verdict is `diverged`, it writes nothing. Follow `## Escalate`.\n6. Run `skillset check --only outputs --root .`. Continue only when it passes.\n\n`--repair` restores a deleted managed file byte-identically and leaves the lock\nuntouched when nothing else changed.\n\n`skillset build --repair --discard-edits` overwrites a hand-edited output instead\nof preserving it. Use it only after the edit has been carried into source, or when\nthe edit is confirmed unwanted. It cannot override a `diverged` verdict.\n\n## Rebuild an untrusted lock\n\nA `lock-untrusted` verdict means the lock's own integrity check failed (its\nheader hash or an entry's hash no longer verifies), so the\nrepair cannot tell a stale output from a hand-edited one. The generated file is\nnot necessarily wrong, and there is probably no misdirected edit to recover.\n\nMUST NOT treat this as `output-edited`. MUST NOT go looking for an edit to port\ninto source.\n\n1. Confirm the generated files themselves are what you expect:\n   `git status` and `git diff` over the named paths.\n2. Rebuild so the lock is regenerated with valid provenance:\n   `skillset build --root . --yes`.\n3. Rerun `skillset build --repair --root .`. The verdicts should now be real.\n4. When the lock was tampered with deliberately, or step 2 does not clear it,\n   stop and report that. A lock that will not rebuild is a compiler problem, not\n   a content problem.\n\n## Recover a misdirected edit\n\nDo this for every `output-edited` path. When the generated file changed and its\nsource did not, the edit was almost certainly aimed at the wrong file. The intent\nis usually right and the target is wrong.\n\nMUST NOT discard the edit. MUST NOT leave the edit in the generated file.\n\n1. Capture the edit: `git diff -- \u003Cgenerated-path>`, or read the backup Skillset\n   wrote under `.skillset\u002Fsnapshots\u002F` before any overwrite.\n2. Find the authoring source: `skillset explain \u003Cgenerated-path> --root .`.\n3. Apply the equivalent change to that source file.\n4. Regenerate: `skillset build --root . --yes`.\n5. Confirm the generated file now contains the intended change, and that\n   `skillset check --only outputs --root .` passes.\n6. When the change cannot be expressed in source, stop and report that. It means\n   either the source format lacks the capability or the change does not belong in\n   generated output.\n\n### Which row is it?\n\n> **`source-ahead`.** A skill's source gained a new section. The generated\n> `SKILL.md` no longer matches what the compiler would emit, and nobody touched the\n> generated file. Regenerate.\n\n> **`output-edited`.** Someone fixed a typo directly in\n> `.agents\u002Fskills\u002F\u003Cname>\u002FSKILL.md`. The source still says what it always said.\n> Port the typo fix to source, then regenerate.\n\n> **`lock-untrusted`.** Nobody touched the generated file at all; a merge or an\n> edit damaged `skillset.lock`, so its recorded hashes no longer verify. There\n> is no edit to port. Rebuild the lock.\n\nBoth look identical in `git status` — one modified generated file. Only the lock\ncomparison separates them, which is why the verdict is worth reading rather than\nguessing.\n\n## Escalate\n\nStop and hand back when either holds:\n\n- A path's verdict is `diverged`: the generated file was edited **and** its source\n  changed.\n- A change cannot be expressed in authored source.\n\nReport the exact paths and which condition applies. MUST NOT guess which side to\nkeep, and MUST NOT resolve the path by picking one.\n\n## Verify before finishing\n\n- Run `skillset check --only outputs --root .`. It MUST pass.\n- Confirm no generated file contains a conflict marker. Generated output merges\n  as ordinary text, so a true conflict writes markers into it; `skillset resolve\n  --yes` regenerates those paths from the index stages, and a marker left\n  behind means a conflicted path was staged without being resolved.\n- Confirm no authored source was staged by a resolve step that should only have\n  touched generated output.\n","# Fix Skillset Generated Output\n\nSkillset commits its generated output. Those files are derived and\nnon-authoritative: their final content must come from authored source, never from\nmanual reconciliation. A conflict carries no valid merged content. A diff can\nstill preserve evidence of a misdirected human edit, so keep it until Skillset\nreports an `output-edited` or `diverged` verdict and the intent is recovered.\n\nNever leave an edit in a generated file, and never hand-merge one.\n\n## Run the repository's own compiler\n\nFirst check whether the repository builds Skillset itself by looking for\n`apps\u002Fskillset\u002Fpackage.json` with package name `skillset`.\n\n- In a Skillset compiler checkout, compare `skillset --version` with the version\n  in `apps\u002Fskillset\u002Fpackage.json`, then invoke the compiler through the repository:\n  `bun .\u002Fapps\u002Fskillset\u002Fsrc\u002Fcli.ts \u003Ccommand>`, or the `bun run skillset:*` scripts.\n- In a consumer repository, use the installed binary. MUST NOT compare it with the\n  root `package.json`; that version belongs to the consuming application.\n\nWhen the compiler-checkout versions differ, a globally installed `skillset` is\nshadowing the one under test, and its answers describe a different compiler.\n\nThis is not hypothetical. A stale 0.26.1 binary reporting `build is blocked` on a\nrepository whose own compiler restores the file correctly is what produced the\nfalse premise behind SET-599.\n\n## Decide whether the path is generated\n\nDo this first. It determines everything that follows.\n\n1. Run `skillset explain \u003Cpath> --root .`. It reports ownership and provenance, and\n   labels a derived path `(generated)`.\n2. When `explain` is unavailable, treat a path as generated if it appears as an\n   `outputPath` or in `files[]` of a `skillset.lock`, or is a `skillset.lock`.\n3. When the path is authored source, this skill does not apply. Resolve it as\n   ordinary source.\n\n## Read the verdict\n\n`skillset build --repair` reports repair verdicts and actions. A dry-run\n`skillset resolve` instead prints the conflict partition (`generated`, authored,\nor hand-edited); after `--yes` reaches the repair preview, it prints any\nnon-clean repair verdicts before its summary. Read the command's exact report\nrather than re-deriving it.\n\n| Verdict | What it means | What happens |\n| --- | --- | --- |\n| `clean` | The file already equals what source produces. | Nothing. |\n| `output-missing` | A managed file is absent. | Restored. No edit can be lost. |\n| `source-ahead` | Source moved; the output is stale. | Regenerated. |\n| `output-obsolete` | The lock claims it but source no longer produces it. | Removed. |\n| `output-edited` | The generated file was edited; source was not. | Preserved, not overwritten. See `## Recover a misdirected edit`. |\n| `lock-untrusted` | The lock records no verdict this repair can rely on. | Preserved. The fault is the lock, not the output. See `## Rebuild an untrusted lock`. |\n| `diverged` | The generated file **and** its source both moved. | Refused. See `## Escalate`. |\n\nThe verdicts come from a three-way comparison against the `outputHash` the lock\nrecorded for the last build — the file on disk, the lock, and what a fresh render\nwould produce.\n\nMUST NOT use file modification times to decide any of this. A rebase or a checkout\nrewrites mtimes across the whole tree, so in the situation where drift matters\nmost, every file appears equally fresh.\n\n`outputHash` is recorded per lock *item*, which is a group of files, so a single\nedited file marks its whole group. A path whose bytes already equal the render is\nreported `clean` regardless, because writing it would change nothing.\n\n## Clear a conflict during a rebase or merge\n\n1. Run `skillset resolve --root .` to see the partition without writing anything.\n2. When it names **authored** conflicts, it stages nothing and exits non-zero.\n   Resolve those files, `git add` them, then return to step 1. Regenerating from\n   source that still carries conflict markers would render those markers into\n   generated output.\n3. When it reports paths **edited by hand on one side of the conflict**, it\n   stages nothing and exits non-zero. Follow `## Recover a misdirected edit` for\n   each named path, run `git add \u003Csource-path>` for the recovered source change,\n   then return to step 1. MUST NOT commit while generated paths are still\n   unmerged; continuation creates the rebased or merged commit after generated\n   output is repaired.\n4. Run `skillset resolve --root . --yes`. It repairs the generated paths and stages\n   them.\n5. When it exits zero, continue the operation with the command that owns it:\n   `gh stack rebase --continue` in a stack, `git rebase --continue` for a rebase, or\n   `git merge --continue` for a merge.\n6. When it reports a `diverged` path, follow `## Escalate`.\n\nA conflicted file on disk represents neither side, so the verdict table has no\nbaseline there. `skillset resolve` instead checks each conflict side against the\nlock that side committed, and refuses when either disagrees. MUST NOT resolve\nsuch a path by taking a side. The pre-rebase commit still holds the edit; recover\nit from there.\n\n`skillset resolve` reads lock ownership from exact index blobs: stage 0 for an\nunconflicted lock and stages 2 and 3 for the two conflict sides. It never trusts\nmutable worktree lock bytes as provenance, so a lock carrying markers does not\nneed fixing by hand first. It stages only paths it classified as generated, and\nMUST NOT be used to stage authored source.\n\nWith `--yes`, resolve refuses worktree-only compiler inputs that could change the\nprojection: source, workspace config and manifest files, external repository\npackage sources, and explicit feature inputs. Staged inputs and unrelated dirt\noutside that input inventory do not block it.\n\nThe repair is transactional across the whole projection, not only the paths that\nhappened to conflict. Before writing, resolve records each affected path's exact\nbytes, executable mode, absence, missing parent directories, or symlink target.\nIf validation, build, or staging fails, it restores those preimages rather than\nleaving a partially repaired worktree.\n\n## Repair drift outside a conflict\n\nUse this when `skillset check` reports drift and no rebase or merge is in progress.\n\n1. Run `skillset build --repair --root .`. It classifies every managed path against\n   filesystem reality and prints a verdict for each that needs action.\n2. When every verdict is `clean`, `output-missing`, `source-ahead`, or\n   `output-obsolete`, run `skillset build --repair --root . --yes`. It restores,\n   regenerates, and removes.\n3. When any verdict is `output-edited`, it writes nothing. Follow\n   `## Recover a misdirected edit` for each named path, then return to step 1.\n4. When any verdict is `lock-untrusted`, it writes nothing. Follow\n   `## Rebuild an untrusted lock`, then return to step 1.\n5. When any verdict is `diverged`, it writes nothing. Follow `## Escalate`.\n6. Run `skillset check --only outputs --root .`. Continue only when it passes.\n\n`--repair` restores a deleted managed file byte-identically and leaves the lock\nuntouched when nothing else changed.\n\n`skillset build --repair --discard-edits` overwrites a hand-edited output instead\nof preserving it. Use it only after the edit has been carried into source, or when\nthe edit is confirmed unwanted. It cannot override a `diverged` verdict.\n\n## Rebuild an untrusted lock\n\nA `lock-untrusted` verdict means the lock's own integrity check failed (its\nheader hash or an entry's hash no longer verifies), so the\nrepair cannot tell a stale output from a hand-edited one. The generated file is\nnot necessarily wrong, and there is probably no misdirected edit to recover.\n\nMUST NOT treat this as `output-edited`. MUST NOT go looking for an edit to port\ninto source.\n\n1. Confirm the generated files themselves are what you expect:\n   `git status` and `git diff` over the named paths.\n2. Rebuild so the lock is regenerated with valid provenance:\n   `skillset build --root . --yes`.\n3. Rerun `skillset build --repair --root .`. The verdicts should now be real.\n4. When the lock was tampered with deliberately, or step 2 does not clear it,\n   stop and report that. A lock that will not rebuild is a compiler problem, not\n   a content problem.\n\n## Recover a misdirected edit\n\nDo this for every `output-edited` path. When the generated file changed and its\nsource did not, the edit was almost certainly aimed at the wrong file. The intent\nis usually right and the target is wrong.\n\nMUST NOT discard the edit. MUST NOT leave the edit in the generated file.\n\n1. Capture the edit: `git diff -- \u003Cgenerated-path>`, or read the backup Skillset\n   wrote under `.skillset\u002Fsnapshots\u002F` before any overwrite.\n2. Find the authoring source: `skillset explain \u003Cgenerated-path> --root .`.\n3. Apply the equivalent change to that source file.\n4. Regenerate: `skillset build --root . --yes`.\n5. Confirm the generated file now contains the intended change, and that\n   `skillset check --only outputs --root .` passes.\n6. When the change cannot be expressed in source, stop and report that. It means\n   either the source format lacks the capability or the change does not belong in\n   generated output.\n\n### Which row is it?\n\n> **`source-ahead`.** A skill's source gained a new section. The generated\n> `SKILL.md` no longer matches what the compiler would emit, and nobody touched the\n> generated file. Regenerate.\n\n> **`output-edited`.** Someone fixed a typo directly in\n> `.agents\u002Fskills\u002F\u003Cname>\u002FSKILL.md`. The source still says what it always said.\n> Port the typo fix to source, then regenerate.\n\n> **`lock-untrusted`.** Nobody touched the generated file at all; a merge or an\n> edit damaged `skillset.lock`, so its recorded hashes no longer verify. There\n> is no edit to port. Rebuild the lock.\n\nBoth look identical in `git status` — one modified generated file. Only the lock\ncomparison separates them, which is why the verdict is worth reading rather than\nguessing.\n\n## Escalate\n\nStop and hand back when either holds:\n\n- A path's verdict is `diverged`: the generated file was edited **and** its source\n  changed.\n- A change cannot be expressed in authored source.\n\nReport the exact paths and which condition applies. MUST NOT guess which side to\nkeep, and MUST NOT resolve the path by picking one.\n\n## Verify before finishing\n\n- Run `skillset check --only outputs --root .`. It MUST pass.\n- Confirm no generated file contains a conflict marker. Generated output merges\n  as ordinary text, so a true conflict writes markers into it; `skillset resolve\n  --yes` regenerates those paths from the index stages, and a marker left\n  behind means a conflicted path was staged without being resolved.\n- Confirm no authored source was staged by a resolve step that should only have\n  touched generated output.\n",{"name":5,"metadata":24,"description":7},{"version":25,"skillset.schema":26},"0.4.0","1","main","24800e02ed4e7150ea1bc7130c35603c98ea9da2",".cursor\u002Fskills\u002Fskillset-fix",0,"https:\u002F\u002Fgithub.com\u002Foutfitter-dev\u002Fskillset","https:\u002F\u002Fgithub.com\u002Foutfitter-dev","https:\u002F\u002Foutfitter.dev","Making things to help agents & devs make things.","2026-11-01 12:23:53",{"source":8,"skillName":5,"commands":37,"prompt":47,"zipAvailable":48},[38,41,44],{"tool":39,"command":40},"npx","npx skills add outfitter-dev\u002Fskillset --skill skillset-fix",{"tool":42,"command":43},"bunx","bunx skills add outfitter-dev\u002Fskillset --skill skillset-fix",{"tool":45,"command":46},"pnpm","pnpm dlx skills add outfitter-dev\u002Fskillset --skill skillset-fix","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Foutfitter-dev\u002Fskillset\u002Fblob\u002Fmain\u002F.cursor\u002Fskills\u002Fskillset-fix\u002FSKILL.md",true,[],[51,59,66,73,80,87],{"skillNo":52,"name":53,"nameZh":6,"description":54,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":55,"sourceHtmlUrl":56,"categories":57,"keywords":58,"publishedAt":19},"01M758DGFEVX68J2QSQHSN08JB","use-skillset","Use the skillset compiler in a repo with skillset.yaml or a .skillset\u002F directory. Covers creating a workspace, adding skills, rules, agents, hooks, and plugins, building and checking Claude, Codex, and Cursor output, repairing stale or hand-edited generated files, recording changes, cutting releases, importing existing skills, and marketplaces. Use when asked to add a rule, record a change, cut a release, or fix generated output.","https:\u002F\u002Fgithub.com\u002Foutfitter-dev\u002Fskillset\u002Fblob\u002Fmain\u002Fplugins\u002Fskillset\u002Fskills\u002Fuse-skillset\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Foutfitter-dev\u002Fskillset\u002Fblob\u002Fmain\u002Fplugins\u002Fskillset\u002Fskills\u002Fuse-skillset\u002FSKILL.md",[],[],{"skillNo":60,"name":61,"nameZh":6,"description":62,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":63,"categories":64,"keywords":65,"publishedAt":19},"01M758DEEVQ3DGGJDRTMJKTPKP","broken","42","https:\u002F\u002Fgithub.com\u002Foutfitter-dev\u002Fskillset\u002Fblob\u002Fmain\u002Ffixtures\u002Fworkbench-invalid\u002F.skillset\u002Fskills\u002Fbroken\u002FSKILL.md",[],[],{"skillNo":67,"name":68,"nameZh":6,"description":69,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":70,"categories":71,"keywords":72,"publishedAt":19},"01M758DCJQYNK9E0DSKPD7JX1S","reference","Keep project references organized for review.","https:\u002F\u002Fgithub.com\u002Foutfitter-dev\u002Fskillset\u002Fblob\u002Fmain\u002Ffixtures\u002Fworkbench-clean\u002F.skillset\u002Fskills\u002Freference\u002FSKILL.md",[],[],{"skillNo":74,"name":75,"nameZh":6,"description":76,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":79,"publishedAt":19},"01M758D9JB3PE4806K3J0TJFJN","portable-proof","Proves that generated Agent Skills are accepted by pinned external consumers.","https:\u002F\u002Fgithub.com\u002Foutfitter-dev\u002Fskillset\u002Fblob\u002Fmain\u002Ffixtures\u002Fstandards-adoption\u002F.skillset\u002Fskills\u002Fportable-proof\u002FSKILL.md",[],[],{"skillNo":81,"name":82,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":19},"01M758D6ZT3PJVNPPZ5RF0BPM6","plugin-proof","Proves immediate-child skill fidelity inside an Agent Plugins package.","https:\u002F\u002Fgithub.com\u002Foutfitter-dev\u002Fskillset\u002Fblob\u002Fmain\u002Ffixtures\u002Fstandards-adoption\u002F.skillset\u002Fplugins\u002Fportable-proof\u002Fskills\u002Fplugin-proof\u002FSKILL.md",[],[],{"skillNo":88,"name":89,"nameZh":6,"description":90,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":91,"categories":92,"keywords":93,"publishedAt":19},"01M758D46M0FV8A6EBG9K6NDN0","sink","Kitchen-sink skill exercising shared resources and custom resource mappings.","https:\u002F\u002Fgithub.com\u002Foutfitter-dev\u002Fskillset\u002Fblob\u002Fmain\u002Ffixtures\u002Fkitchen-sink\u002F.skillset\u002Fplugins\u002Fkitchen\u002Fskills\u002Fsink\u002FSKILL.md",[],[]]