[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7D3E304MTYQNNGYX0X2PGSW":3,"skill-same:01M7D3E304MTYQNNGYX0X2PGSW":57},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":25,"commitSha":26,"skillDirectory":27,"forksCount":28,"repositoryHtmlUrl":29,"repositoryHomepageUrl":6,"ownerHtmlUrl":30,"ownerBlogUrl":31,"ownerDescription":6,"updatedAt":32,"install":33,"resources":46},"01M7D3E304MTYQNNGYX0X2PGSW","writing-skills",null,"Human-started command: it runs only when the human's message begins with `\u002Fak:writing-skills`. On any other request do not load or follow it; tell the human to type that command. Authors a new skill or edits an existing one in this package, test-first: a failing eval case and a recorded baseline come before any wording, and the change stops as a candidate for a separate review. Use when a human asks to write, change or harden a skill, protocol, role or reference in this catalog. Not for project documentation, and not for promoting a change.","Pibomeister\u002Fagent-kit","Pibomeister","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F11633791?v=4","01M7D2NBHEKARMHN9MPV2J80XD","Eduardo Picazo","01M7D2NBJ3Z33SGSYCZ0KBD5DE",0,"MIT","https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fwriting-skills\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: writing-skills\ndescription: >-\n  Human-started command: it runs only when the human's message begins with `\u002Fak:writing-skills`. On\n  any other request do not load or follow it; tell the human to type that command. Authors a new\n  skill or edits an existing one in this package, test-first: a failing eval case and a recorded\n  baseline come before any wording, and the change stops as a candidate for a separate review. Use\n  when a human asks to write, change or harden a skill, protocol, role or reference in this catalog.\n  Not for project documentation, and not for promoting a change.\nlicense: MIT\nmetadata:\n  ak_catalog_id: writing-skills\n---\n\nAuthor and improve skills using behavioral tests and provenance. Required for maintaining this\npackage; candidate edits pass evaluations and a separate review before promotion.\n\n## When to use\n\n- A human asks to write a new skill, protocol, role or reference for this catalog.\n- A human asks to change an existing one: a gate that does not hold, a wrong-shaped output, a\n  missing element.\n- A supported lesson in the knowledgebase names a skill's instructions as the cause of a failure,\n  and the human asks to turn it into a candidate change.\n- A revision failed its evaluations or its review, and the human asks to roll it back.\n\n## Not for\n\n- Project documentation, decisions or lessons. Those belong to the central knowledgebase and to\n  `compound`, and this skill writes none of them into the repository.\n- Promoting a change. The author of a candidate never promotes it; evaluations and a separate\n  review do.\n- Fixing a defect in the `ak` CLI or its validators. That is ordinary code under `src\u002F`, done\n  test-first under the `tdd` protocol.\n- Rewording a document for readability alone. That is the prose-quality reference pack, applied by\n  whichever skill owns the document.\n\n## Authority\n\nAuthority: `explicit`. A human starts it with `\u002Fak:writing-skills`. No phase operation exposes it\n(`policies\u002Finvocation.yaml` lists it among the user-invoked skills that expose none), so no\ncontroller can start it under a grant. Started any other way, it stops at step 1 and names the\ncommand.\n\n## Inputs\n\n- The target: a skill, protocol, role or reference id in `catalog.yaml`. A new id with no entry\n  there: `needs-input` for the catalog owner, who adds the entry first.\n- The failure to fix: an observed transcript, a failing eval case, or a knowledgebase lesson read\n  through `readContext`. None of these: the baseline in step 3 must produce one.\n- The eval runner and the repository at a readable revision. Unavailable: `needs-input`.\n\n## Workflow\n\n1. **Check how this run was started**, before any other step and before any tool call. It is started\n   only when the human's message begins with `\u002Fak:writing-skills`; no grant starts it. A request in\n   prose is not a start, even when it names this skill or the command. Otherwise, stop here: make no\n   tool call, say that this command is human-started, and give the human the line to type,\n   `\u002Fak:writing-skills` and their request.\n2. **Read the contract.** Read `AGENTS.md`, `AUTHORING.md`, the target's `catalog.yaml` entry, the\n   rulings that bind it in `policies\u002Fresolved-conflicts.yaml`, and the current output of\n   `bun run ak validate`. They govern; this skill does not restate them. Where they are silent or\n   wrong, file a `CONTRACT-DEFECTS.md` entry and stop on that item.\n3. **RED: watch it fail.** Pick the eval case that exercises the failure, or write one under\n   `evals\u002F\u003Cid>\u002F`. Run it without the change, in a context that has not seen the change. Record the\n   behavior and every rationalization verbatim. If the baseline does not fail, stop: there is\n   nothing to fix.\n4. **Classify the failure and choose the form.** Load\n   [the authoring-discipline reference](references\u002Fauthoring-discipline.md) and match the form to\n   the failure. A rationalization table is for discipline failures only, and its rows come from\n   the rationalizations recorded in step 3.\n5. **GREEN: write the smallest change.** Load\n   [the prose-quality reference pack](..\u002F..\u002Freferences\u002Fprose-quality\u002FREFERENCE.md) and write to it.\n   Change the body and its `skill.yaml` mirror together, add a provenance row for every adapted\n   file, and cite each binding ruling where the contract says to.\n6. **Micro-test the wording when a full run is costly.** Use the protocol in the\n   authoring-discipline reference: a no-guidance control, five or more samples per variant, every\n   match read by hand. Micro-tests check wording; they never replace step 7.\n7. **Verify.** Run the same cases with the change, in a fresh context. Then run\n   `bun run ak validate` and `bun run test`. Record each result with the revision it measured.\n8. **REFACTOR: close loopholes.** A new rationalization in step 7 gets an explicit counter and a\n   re-run. Repeat until the cases hold.\n9. **Publish the change record and stop.** Publish the record through `publishArtifact`, then read\n   it back. Return the change as a candidate for a separate review. Do not promote it.\n\nRolling back replaces steps 3–8: revert the target's source to its last verified revision, run\n`bun run ak validate` and `bun run test`, then publish a change record that notes the rollback (step 9).\nLessons and earlier change records are left as they are.\n\n## Hard gates\n\nGate: no wording before a failing case. This holds for new skills and for edits to existing ones.\nA change written first is discarded, not kept as a reference while the case is written. A rollback\nrestores verified wording rather than writing new wording, so step 3 does not apply to it.\n\nGate: no promotion without the recorded baseline, the verified run and a separate review. The\nauthor of the change is never its reviewer, and never promotes it.\n\nGate: a rationalization table only for a discipline failure, with rows taken from the baseline.\nShape and omission failures get a recipe or a required slot.\n\nGate: rolling a skill back reverts skill source only. The lesson and the change record that\nmotivated it stay in the knowledgebase with their history.\n\nGate: nothing project-derived is written into this repository. Skill source, eval cases and\nprovenance rows are this package's own reusable instructions; the change record is not (ruling\n`central-kb-owns-project-artifacts`).\n\n| The thought | Why it is wrong | Do this instead |\n|---|---|---|\n| \"The skill is obviously clear, so it does not need a test.\" | Clear to its author is not clear to the agent that runs it. | Run the baseline and watch it fail first. |\n| \"It is only a small addition to one section.\" | A small edit is still an edit, and the gate covers edits. | Write or pick the failing case before the wording. |\n| \"It passed once; that is enough to ship.\" | A single sample does not show the wording binds, and passing is not review. | Record the run, and hand the candidate to a separate reviewer. |\n| \"I will add a rationalization table to be safe.\" | On a shape or omission failure, prohibitions add the unwanted content. | Classify the failure; use a recipe or a required slot. |\n| \"The skill reverted, so its lesson should go too.\" | The lesson's evidence did not change when the wording did. | Revert the source; leave the knowledge history intact. |\n\n## Outputs\n\n- A candidate change to this package: skill source, its `skill.yaml`, eval cases and provenance\n  rows. It is not promoted.\n- A change record: the target, the baseline and its verbatim rationalizations, the form chosen and\n  why, and the verified results with the revision each measured. It is published as a knowledgebase\n  `process` document for this catalog through `publishArtifact`, never to a repository path (ruling\n  `central-kb-owns-project-artifacts`; `docs\u002Fdecisions\u002F0001-kb-document-vocabulary.md` §3).\n- With the baseline passing: no change, and the baseline result returned.\n\n## Side effects\n\n`skill-source-write`, `process-exec`, `artifact-write`, `scratch-write`, `kb-publish`. No\n`workspace-write`: the only repository files this skill changes are this package's own skill\nsource, eval cases and provenance rows, and nothing project-derived is written here (ruling\n`central-kb-owns-project-artifacts`).\n\n`kb-publish` is a remote side effect. Its idempotency key is the one\n`adapters\u002Frunner-contract\u002FCONTRACT.md` §5 derives, with the input artifact hash taken over the\ndraft persisted in scratch, which a resumed run reuses. The read-back is the record as the\nknowledgebase returns it. A resumed run reads back first and does not publish a second copy. A\npublish whose read-back cannot be performed is\n`failed`, never complete.\n\n## Stop conditions\n\n- `complete` — the candidate is verified, the change record is published and read back, and the\n  candidate is returned for a separate review.\n- `complete` — the baseline did not fail; nothing was changed, and the result is returned.\n- `complete` — a rollback is verified, and its change record is published and read back.\n- `needs-input` — started without an explicit `\u002Fak:writing-skills`; or the target has no\n  `catalog.yaml` entry, for the catalog owner to add; or the repository or eval runner is\n  unavailable; or `kb-write` is unavailable, and the change record is returned in the session\n  instead; or the contract is silent or wrong and a defect entry was filed.\n- `failed` — the change record's read-back could not be performed.\n\n## Limits\n\n- Changes written before a failing case: 0 (gate).\n","Author and improve skills using behavioral tests and provenance. Required for maintaining this\npackage; candidate edits pass evaluations and a separate review before promotion.\n\n## When to use\n\n- A human asks to write a new skill, protocol, role or reference for this catalog.\n- A human asks to change an existing one: a gate that does not hold, a wrong-shaped output, a\n  missing element.\n- A supported lesson in the knowledgebase names a skill's instructions as the cause of a failure,\n  and the human asks to turn it into a candidate change.\n- A revision failed its evaluations or its review, and the human asks to roll it back.\n\n## Not for\n\n- Project documentation, decisions or lessons. Those belong to the central knowledgebase and to\n  `compound`, and this skill writes none of them into the repository.\n- Promoting a change. The author of a candidate never promotes it; evaluations and a separate\n  review do.\n- Fixing a defect in the `ak` CLI or its validators. That is ordinary code under `src\u002F`, done\n  test-first under the `tdd` protocol.\n- Rewording a document for readability alone. That is the prose-quality reference pack, applied by\n  whichever skill owns the document.\n\n## Authority\n\nAuthority: `explicit`. A human starts it with `\u002Fak:writing-skills`. No phase operation exposes it\n(`policies\u002Finvocation.yaml` lists it among the user-invoked skills that expose none), so no\ncontroller can start it under a grant. Started any other way, it stops at step 1 and names the\ncommand.\n\n## Inputs\n\n- The target: a skill, protocol, role or reference id in `catalog.yaml`. A new id with no entry\n  there: `needs-input` for the catalog owner, who adds the entry first.\n- The failure to fix: an observed transcript, a failing eval case, or a knowledgebase lesson read\n  through `readContext`. None of these: the baseline in step 3 must produce one.\n- The eval runner and the repository at a readable revision. Unavailable: `needs-input`.\n\n## Workflow\n\n1. **Check how this run was started**, before any other step and before any tool call. It is started\n   only when the human's message begins with `\u002Fak:writing-skills`; no grant starts it. A request in\n   prose is not a start, even when it names this skill or the command. Otherwise, stop here: make no\n   tool call, say that this command is human-started, and give the human the line to type,\n   `\u002Fak:writing-skills` and their request.\n2. **Read the contract.** Read `AGENTS.md`, `AUTHORING.md`, the target's `catalog.yaml` entry, the\n   rulings that bind it in `policies\u002Fresolved-conflicts.yaml`, and the current output of\n   `bun run ak validate`. They govern; this skill does not restate them. Where they are silent or\n   wrong, file a `CONTRACT-DEFECTS.md` entry and stop on that item.\n3. **RED: watch it fail.** Pick the eval case that exercises the failure, or write one under\n   `evals\u002F\u003Cid>\u002F`. Run it without the change, in a context that has not seen the change. Record the\n   behavior and every rationalization verbatim. If the baseline does not fail, stop: there is\n   nothing to fix.\n4. **Classify the failure and choose the form.** Load\n   [the authoring-discipline reference](references\u002Fauthoring-discipline.md) and match the form to\n   the failure. A rationalization table is for discipline failures only, and its rows come from\n   the rationalizations recorded in step 3.\n5. **GREEN: write the smallest change.** Load\n   [the prose-quality reference pack](..\u002F..\u002Freferences\u002Fprose-quality\u002FREFERENCE.md) and write to it.\n   Change the body and its `skill.yaml` mirror together, add a provenance row for every adapted\n   file, and cite each binding ruling where the contract says to.\n6. **Micro-test the wording when a full run is costly.** Use the protocol in the\n   authoring-discipline reference: a no-guidance control, five or more samples per variant, every\n   match read by hand. Micro-tests check wording; they never replace step 7.\n7. **Verify.** Run the same cases with the change, in a fresh context. Then run\n   `bun run ak validate` and `bun run test`. Record each result with the revision it measured.\n8. **REFACTOR: close loopholes.** A new rationalization in step 7 gets an explicit counter and a\n   re-run. Repeat until the cases hold.\n9. **Publish the change record and stop.** Publish the record through `publishArtifact`, then read\n   it back. Return the change as a candidate for a separate review. Do not promote it.\n\nRolling back replaces steps 3–8: revert the target's source to its last verified revision, run\n`bun run ak validate` and `bun run test`, then publish a change record that notes the rollback (step 9).\nLessons and earlier change records are left as they are.\n\n## Hard gates\n\nGate: no wording before a failing case. This holds for new skills and for edits to existing ones.\nA change written first is discarded, not kept as a reference while the case is written. A rollback\nrestores verified wording rather than writing new wording, so step 3 does not apply to it.\n\nGate: no promotion without the recorded baseline, the verified run and a separate review. The\nauthor of the change is never its reviewer, and never promotes it.\n\nGate: a rationalization table only for a discipline failure, with rows taken from the baseline.\nShape and omission failures get a recipe or a required slot.\n\nGate: rolling a skill back reverts skill source only. The lesson and the change record that\nmotivated it stay in the knowledgebase with their history.\n\nGate: nothing project-derived is written into this repository. Skill source, eval cases and\nprovenance rows are this package's own reusable instructions; the change record is not (ruling\n`central-kb-owns-project-artifacts`).\n\n| The thought | Why it is wrong | Do this instead |\n|---|---|---|\n| \"The skill is obviously clear, so it does not need a test.\" | Clear to its author is not clear to the agent that runs it. | Run the baseline and watch it fail first. |\n| \"It is only a small addition to one section.\" | A small edit is still an edit, and the gate covers edits. | Write or pick the failing case before the wording. |\n| \"It passed once; that is enough to ship.\" | A single sample does not show the wording binds, and passing is not review. | Record the run, and hand the candidate to a separate reviewer. |\n| \"I will add a rationalization table to be safe.\" | On a shape or omission failure, prohibitions add the unwanted content. | Classify the failure; use a recipe or a required slot. |\n| \"The skill reverted, so its lesson should go too.\" | The lesson's evidence did not change when the wording did. | Revert the source; leave the knowledge history intact. |\n\n## Outputs\n\n- A candidate change to this package: skill source, its `skill.yaml`, eval cases and provenance\n  rows. It is not promoted.\n- A change record: the target, the baseline and its verbatim rationalizations, the form chosen and\n  why, and the verified results with the revision each measured. It is published as a knowledgebase\n  `process` document for this catalog through `publishArtifact`, never to a repository path (ruling\n  `central-kb-owns-project-artifacts`; `docs\u002Fdecisions\u002F0001-kb-document-vocabulary.md` §3).\n- With the baseline passing: no change, and the baseline result returned.\n\n## Side effects\n\n`skill-source-write`, `process-exec`, `artifact-write`, `scratch-write`, `kb-publish`. No\n`workspace-write`: the only repository files this skill changes are this package's own skill\nsource, eval cases and provenance rows, and nothing project-derived is written here (ruling\n`central-kb-owns-project-artifacts`).\n\n`kb-publish` is a remote side effect. Its idempotency key is the one\n`adapters\u002Frunner-contract\u002FCONTRACT.md` §5 derives, with the input artifact hash taken over the\ndraft persisted in scratch, which a resumed run reuses. The read-back is the record as the\nknowledgebase returns it. A resumed run reads back first and does not publish a second copy. A\npublish whose read-back cannot be performed is\n`failed`, never complete.\n\n## Stop conditions\n\n- `complete` — the candidate is verified, the change record is published and read back, and the\n  candidate is returned for a separate review.\n- `complete` — the baseline did not fail; nothing was changed, and the result is returned.\n- `complete` — a rollback is verified, and its change record is published and read back.\n- `needs-input` — started without an explicit `\u002Fak:writing-skills`; or the target has no\n  `catalog.yaml` entry, for the catalog owner to add; or the repository or eval runner is\n  unavailable; or `kb-write` is unavailable, and the change record is returned in the session\n  instead; or the contract is silent or wrong and a defect entry was filed.\n- `failed` — the change record's read-back could not be performed.\n\n## Limits\n\n- Changes written before a failing case: 0 (gate).\n",{"name":5,"license":15,"metadata":24,"description":7},{"ak_catalog_id":5},"main","e0232821c4abbcd4338d6018c20fb70f1b74336b","skills\u002Fwriting-skills",3,"https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit","https:\u002F\u002Fgithub.com\u002FPibomeister","www.eduardopicazoenriquez.com","2026-11-15 00:56:58",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add Pibomeister\u002Fagent-kit --skill writing-skills",{"tool":39,"command":40},"bunx","bunx skills add Pibomeister\u002Fagent-kit --skill writing-skills",{"tool":42,"command":43},"pnpm","pnpm dlx skills add Pibomeister\u002Fagent-kit --skill writing-skills","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fwriting-skills\u002FSKILL.md",true,[47,53],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"references\u002Fauthoring-discipline.md","REFERENCE","参考文档",7483,"https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fwriting-skills\u002Freferences\u002Fauthoring-discipline.md",{"path":54,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":55,"sourceHtmlUrl":56},"skill.yaml",11983,"https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fwriting-skills\u002Fskill.yaml",[58,65,70,75,80,85],{"skillNo":59,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":62,"categories":63,"keywords":64,"publishedAt":19},"01M7D3JZP2026RGY6XX1MC3NQQ","alpha","Human-started command: it runs only when the human's message begins with `\u002Fak:alpha`. On any other request do not load or follow it; tell the human to type that command. Runs the alpha workflow when a human asks for it.","https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fvalid\u002Fminimal\u002Fskills\u002Falpha\u002FSKILL.md",[],[],{"skillNo":66,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":19},"01M7D3JA593Q3RBYJXKVZ37Y2K","https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Finvalid\u002F25-article-tier-outside-vocabulary\u002Fskills\u002Falpha\u002FSKILL.md",[],[],{"skillNo":71,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"publishedAt":19},"01M7D3HQDDAZTAPGWCPGDPNVBQ","https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Finvalid\u002F24-article-without-triggers\u002Fskills\u002Falpha\u002FSKILL.md",[],[],{"skillNo":76,"name":60,"nameZh":6,"description":61,"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},"01M7D3HK074TK3ZWK1ZZNQVPB0","https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Finvalid\u002F23-enforced-by-without-predicate\u002Fskills\u002Falpha\u002FSKILL.md",[],[],{"skillNo":81,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":82,"categories":83,"keywords":84,"publishedAt":19},"01M7D3HGHQM5C5TBSXBCE2WFHW","https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Finvalid\u002F22-article-without-source-span\u002Fskills\u002Falpha\u002FSKILL.md",[],[],{"skillNo":86,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":87,"categories":88,"keywords":89,"publishedAt":19},"01M7D3HDV6HMK2ZHRM0P368KNT","https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Finvalid\u002F21-fragment-redeclares-id\u002Fskills\u002Falpha\u002FSKILL.md",[],[]]