[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7353502JYV34164RSC6HKTC":3,"skill-same:01M7353502JYV34164RSC6HKTC":54},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":33,"commitSha":34,"skillDirectory":35,"forksCount":36,"repositoryHtmlUrl":37,"repositoryHomepageUrl":38,"ownerHtmlUrl":39,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":40,"install":41,"resources":53},"01M7353502JYV34164RSC6HKTC","fact-check-prose",null,"Verify claims and values in prose.","Morrison-Lab\u002Fai-config","Morrison-Lab","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F104235520?v=4","01M734G0451S2G35MN4HFQK7XP","Morrison Lab","01M734G059BYFDFQA8YTPDGPMH",4,"MIT","https:\u002F\u002Fgithub.com\u002FMorrison-Lab\u002Fai-config\u002Fblob\u002Fmain\u002Fskills\u002Ffact-check-prose\u002FSKILL.md",[],[],"2026-11-11 15:11:14",false,"---\nname: fact-check-prose\ndescription: \"Verify claims and values in prose.\"\nuser-invocable: true\nallowed-tools:\n  - Bash\n  - Read\n  - Grep\n  - Glob\n  - WebFetch\n  - Edit\n  - Write\n---\n\n# fact-check-prose — verify claims, reasoning, and computed values in prose\n\nProse review usually stops at style. This skill goes further: it checks\nwhether the prose is actually **true** and its **logic actually holds** —\nper the standing rule in `Morrison-Lab\u002Fai-config`'s\n[`shared\u002Fwriting\u002Ffact-check-prose.md`](..\u002F..\u002Fshared\u002Fwriting\u002Ffact-check-prose.md),\nwhich is the canonical statement of this policy (edit that file, not this\nskill's procedure, if the policy itself changes).\n\n## When this fires\n\n- \"fact-check this\", \"check the math\", \"verify this reasoning\", \"check this\n  proof\", \"is this accurate\", \"does this claim check out\"\n- As part of any PR\u002FMR review where the diff touches narrative prose — docs,\n  READMEs, lecture notes, textbook chapters, commit\u002FPR descriptions making a\n  substantive claim. Run it alongside (not instead of) style review\n  (`use-preferred-style`, `find-ai-tells`) and terminology review\n  (`challenge-ambiguous-terminology`, folded into `ard`\u002F`ardi`).\n\n## Procedure\n\n### 1. Identify the target and extract checkable units\n\nA PR\u002FMR diff (`gh pr diff \u003CN>` \u002F `glab mr diff \u003CN>`), a file, or pasted text.\nRead the changed prose and pull out the checkable units:\n\n- **Factual claims** — statements that cite a value, a behavior, a result, an\n  external fact (\"X was introduced in version Y\", \"the estimator is\n  unbiased under Z\").\n- **Undefended claims** — a factual claim with no internal reasoning and no\n  citation backing it, even if it turns out to be accurate.\n  Flag it\n  separately from an accuracy check (below) — a claim can pass accuracy and\n  still fail this one.\n- **Reasoning chains** — an argument the document makes, not just its\n  individual facts.\n  This includes **formal mathematical reasoning**\n  (derivations, proofs, algebraic manipulations — check every step follows\n  from the last, check dimensional\u002Funit consistency, check edge cases) and\n  **informal reasoning** (X therefore Y, a causal claim, a design\n  justification).\n- **Computed\u002Frendered values** — a number, table entry, or figure the prose\n  describes as the output of code, a model fit, or a render.\n- **Citations to external works** --- identify all sources cited that will need to be checked.\n- **Config list and allowlist criteria** --- when prose describes why an entry\n  belongs on an allowlist or exclusion list, or what its entries have in common,\n  verify the claim against the defining comment or documentation in code.\n- **Protections, sandboxes, and security controls** --- when prose documents\n  how to configure a protection (a sandbox, an allowlist, a permission, a firewall),\n  check each mode, platform limitation, precedence rule, and escape hatch against\n  source documentation before pushing.\n\n### 2. Pre-flight Readiness Gate (Hard Stop)\n\nBefore evaluating any claim against an external source,\ncheck whether the repository's local source corpus (`sources\u002F` or equivalent)\nactually contains the needed sources:\n**(Note: If the target document contains zero citations to external works, skip this gate and proceed to Step 3).**\n\n1. **Are required sources missing from the local corpus?**\n2. **Is the required source present but unconverted?**\n   For example, a `.pdf`, `.docx`, or `.epub` file without a corresponding Markdown conversion.\n\nIf **either** of these conditions hold, **refuse to run**.\n- Do NOT silently narrow scope.\n- Do NOT partially check around the gap.\n- Do NOT fall back to memory or the open web for the missing sources.\n\nReturn **`Pre-flight: NOT READY`** instead of a report,\nand print remediation instructions.\nFor example, \"Populate the `sources\u002F` directory with Markdown conversions of the cited works so they can be verified locally.\"\n\nOnly proceed to Step 3 if the gate passes.\n\n### 3. Check factual claims\n\nFor each factual claim, check it against domain knowledge first.\nWhere it's\ncheckable against an external source --- a package's own docs, a spec, a\npaper, a dataset's documentation --- fetch that source **from the local `sources\u002F` corpus**.\nDo **not** fall back to the open web (`WebFetch`) if the local source is missing.\nIf you discover during verification that a local source is incomplete (e.g. missing a chapter), abort and return **`Pre-flight: NOT READY`**.\nConfirm or refute the claim against the local source, not against a secondhand summary.\nDon't wave a plausible-sounding claim through\nunchecked — see\n[`challenge-ambiguous-terminology`](..\u002F..\u002Fshared\u002Fworkflow\u002Fchallenge-ambiguous-terminology.md)\nfor the sibling rule on ambiguous phrasing specifically.\n\nSeparately from accuracy, check that the claim is **defended**: does the\nsurrounding text give reasoning for it, or does it carry a citation? A\nclaim can be accurate and still undefended — flag both kinds of gap, not\njust outright inaccuracy.\n\nWhen prose documents how to configure a protection or security control,\nverify four specific dimensions against the upstream source docs:\n- **Modes:** what happens in non-default modes (e.g. bypass vs prompt vs strict).\n- **Platforms:** where it does not run, and whether it fails open or fails closed.\n- **Precedence:** what overrides it (project vs user settings, flags, env vars),\n  and whether list-valued options merge or replace.\n- **Escape hatches:** retries, exclusions, exceptions, or unmonitored tools.\n\n### 4. Verify document-internal reasoning\n\nWalk each reasoning chain step by step:\n\n- **Formal math** — re-derive or spot-check each step; confirm dimensions and\n  units are consistent at every operation; check boundary\u002Fedge cases the\n  claim implies; confirm a cited theorem's hypotheses actually hold in this\n  application (e.g. don't apply a result that requires independence to\n  correlated data without noting it).\n- **Informal arguments** — does the stated reasoning actually support the\n  conclusion drawn, or is there a gap (an unstated assumption doing the\n  real work, a non sequitur, an overgeneralization from one example)?\n\n### 5. Cross-check computed values and figures against the rendered output\n\nDon't trust the source prose's own description of a computed value — verify\nit against what was actually produced:\n\n1. Check whether the project has a **PR-preview deploy**. Look for a sticky\n   preview-link comment on the PR, or check the project's `CLAUDE.md` \u002F\n   `.github\u002Fworkflows\u002F` for a preview workflow (e.g. `preview.yml` +\n   `preview-deploy.yml` built on `rossjrw\u002Fpr-preview-action`, common in this\n   lab's Quarto repos — those deploy to a `gh-pages` branch under\n   `pr-preview\u002Fpr-\u003CN>\u002F\u003Cpath>`).\n2. If a live preview exists, fetch the rendered page (`WebFetch`) and read\n   the actual value\u002Ffigure off it.\n3. If no preview covers the changed page yet, fall back to the project's\n   render command (e.g. `quarto render \u003Cchapter.qmd> --to html` in an R\u002F\n   Quarto repo) and check the local render instead of skipping the check.\n4. Compare what's rendered against what the prose claims. A mismatch is a\n   finding even if the prose \"sounds\" right — the rendered output is ground\n   truth, not the sentence describing it.\n\n### 6. Report\n\nFor every unit checked:\n\n```\n| Claim\u002Fstep | Verdict | Basis (source checked) |\n|------------|---------|-------------------------|\n| \"the estimator is unbiased\" | Inaccurate | Biased under this design per [source\u002Fderivation]; see step 3 below |\n| Derivation step 4 → 5 | Confirmed | Re-derived; algebra holds, dimensions consistent |\n| \"Table 2, row 3 = 4.7\" | Inaccurate | Rendered preview (pr-preview\u002Fpr-N\u002Fchapter.html) shows 4.9 |\n| \"adoption has grown steadily\" | Undefended | No citation or reasoning given; claim itself checks out but needs a source or an explicit basis |\n```\n\nThen, separately, list **additional citations\u002Freferences worth adding** —\nproactively, anywhere a pointer to a source would help a reader, not only\nwhere a claim is currently flagged as uncited.\n\n### 7. Offer to apply corrections\n\nOn request, fix the flagged inaccuracies and reasoning gaps in place with\n`Edit`, preserving the author's intent and voice — same posture as\n`find-ai-tells`'s apply step. Don't silently rewrite without being asked;\nreport first.\n\n## Relationship to other skills\n\n- [`challenge-ambiguous-terminology`](..\u002F..\u002Fshared\u002Fworkflow\u002Fchallenge-ambiguous-terminology.md) —\n  catches phrasing whose meaning is *unresolved*; this skill catches claims\n  and reasoning that are resolved but *wrong*. Run both during review.\n- **`use-preferred-style` \u002F `find-ai-tells`** — style and AI-tell scans; this\n  skill is the accuracy\u002Freasoning counterpart, not a replacement.\n- **`ard` \u002F `ardi`** — when a reviewer (human or bot) flags a factual or\n  reasoning error, `ard` handles the response disposition; this skill is how\n  you (as reviewer) *find* that class of error in the first place.\n- **`prose-fact-checker` agent** — the read-only, parallelizable version of\n  step 2–4 above, for fanning claim-checking out across a `Workflow` when a\n  document has many independent claims to verify at once.\n- **`check-info-quality` (`ciq`)** — this skill verdicts a claim\n  **Accurate \u002F Inaccurate \u002F Undefended**; `ciq`'s check C catches a claim\n  that's accurate by this skill's own standard yet still misleads (missing\n  context, cherry-picking, a citation that only weakly supports it). `ciq`\n  also covers two failure modes this skill doesn't: stale claims (check A)\n  and irrelevant content (check B). Run both.\n\n## Anti-patterns\n\n- **Don't** accept a claim because it sounds authoritative — check it.\n- **Don't** skip a derivation step because \"it's probably fine\" — re-derive\n  it or flag it as unverified, not silently pass it.\n- **Don't** trust the prose's own description of a computed value when a\n  rendered artifact (preview or local render) is available to check it\n  against directly.\n- **Don't** limit citation suggestions to claims already flagged as uncited —\n  suggest them anywhere they'd help a reader, per the fragment's policy.\n","# fact-check-prose — verify claims, reasoning, and computed values in prose\n\nProse review usually stops at style. This skill goes further: it checks\nwhether the prose is actually **true** and its **logic actually holds** —\nper the standing rule in `Morrison-Lab\u002Fai-config`'s\n[`shared\u002Fwriting\u002Ffact-check-prose.md`](..\u002F..\u002Fshared\u002Fwriting\u002Ffact-check-prose.md),\nwhich is the canonical statement of this policy (edit that file, not this\nskill's procedure, if the policy itself changes).\n\n## When this fires\n\n- \"fact-check this\", \"check the math\", \"verify this reasoning\", \"check this\n  proof\", \"is this accurate\", \"does this claim check out\"\n- As part of any PR\u002FMR review where the diff touches narrative prose — docs,\n  READMEs, lecture notes, textbook chapters, commit\u002FPR descriptions making a\n  substantive claim. Run it alongside (not instead of) style review\n  (`use-preferred-style`, `find-ai-tells`) and terminology review\n  (`challenge-ambiguous-terminology`, folded into `ard`\u002F`ardi`).\n\n## Procedure\n\n### 1. Identify the target and extract checkable units\n\nA PR\u002FMR diff (`gh pr diff \u003CN>` \u002F `glab mr diff \u003CN>`), a file, or pasted text.\nRead the changed prose and pull out the checkable units:\n\n- **Factual claims** — statements that cite a value, a behavior, a result, an\n  external fact (\"X was introduced in version Y\", \"the estimator is\n  unbiased under Z\").\n- **Undefended claims** — a factual claim with no internal reasoning and no\n  citation backing it, even if it turns out to be accurate.\n  Flag it\n  separately from an accuracy check (below) — a claim can pass accuracy and\n  still fail this one.\n- **Reasoning chains** — an argument the document makes, not just its\n  individual facts.\n  This includes **formal mathematical reasoning**\n  (derivations, proofs, algebraic manipulations — check every step follows\n  from the last, check dimensional\u002Funit consistency, check edge cases) and\n  **informal reasoning** (X therefore Y, a causal claim, a design\n  justification).\n- **Computed\u002Frendered values** — a number, table entry, or figure the prose\n  describes as the output of code, a model fit, or a render.\n- **Citations to external works** --- identify all sources cited that will need to be checked.\n- **Config list and allowlist criteria** --- when prose describes why an entry\n  belongs on an allowlist or exclusion list, or what its entries have in common,\n  verify the claim against the defining comment or documentation in code.\n- **Protections, sandboxes, and security controls** --- when prose documents\n  how to configure a protection (a sandbox, an allowlist, a permission, a firewall),\n  check each mode, platform limitation, precedence rule, and escape hatch against\n  source documentation before pushing.\n\n### 2. Pre-flight Readiness Gate (Hard Stop)\n\nBefore evaluating any claim against an external source,\ncheck whether the repository's local source corpus (`sources\u002F` or equivalent)\nactually contains the needed sources:\n**(Note: If the target document contains zero citations to external works, skip this gate and proceed to Step 3).**\n\n1. **Are required sources missing from the local corpus?**\n2. **Is the required source present but unconverted?**\n   For example, a `.pdf`, `.docx`, or `.epub` file without a corresponding Markdown conversion.\n\nIf **either** of these conditions hold, **refuse to run**.\n- Do NOT silently narrow scope.\n- Do NOT partially check around the gap.\n- Do NOT fall back to memory or the open web for the missing sources.\n\nReturn **`Pre-flight: NOT READY`** instead of a report,\nand print remediation instructions.\nFor example, \"Populate the `sources\u002F` directory with Markdown conversions of the cited works so they can be verified locally.\"\n\nOnly proceed to Step 3 if the gate passes.\n\n### 3. Check factual claims\n\nFor each factual claim, check it against domain knowledge first.\nWhere it's\ncheckable against an external source --- a package's own docs, a spec, a\npaper, a dataset's documentation --- fetch that source **from the local `sources\u002F` corpus**.\nDo **not** fall back to the open web (`WebFetch`) if the local source is missing.\nIf you discover during verification that a local source is incomplete (e.g. missing a chapter), abort and return **`Pre-flight: NOT READY`**.\nConfirm or refute the claim against the local source, not against a secondhand summary.\nDon't wave a plausible-sounding claim through\nunchecked — see\n[`challenge-ambiguous-terminology`](..\u002F..\u002Fshared\u002Fworkflow\u002Fchallenge-ambiguous-terminology.md)\nfor the sibling rule on ambiguous phrasing specifically.\n\nSeparately from accuracy, check that the claim is **defended**: does the\nsurrounding text give reasoning for it, or does it carry a citation? A\nclaim can be accurate and still undefended — flag both kinds of gap, not\njust outright inaccuracy.\n\nWhen prose documents how to configure a protection or security control,\nverify four specific dimensions against the upstream source docs:\n- **Modes:** what happens in non-default modes (e.g. bypass vs prompt vs strict).\n- **Platforms:** where it does not run, and whether it fails open or fails closed.\n- **Precedence:** what overrides it (project vs user settings, flags, env vars),\n  and whether list-valued options merge or replace.\n- **Escape hatches:** retries, exclusions, exceptions, or unmonitored tools.\n\n### 4. Verify document-internal reasoning\n\nWalk each reasoning chain step by step:\n\n- **Formal math** — re-derive or spot-check each step; confirm dimensions and\n  units are consistent at every operation; check boundary\u002Fedge cases the\n  claim implies; confirm a cited theorem's hypotheses actually hold in this\n  application (e.g. don't apply a result that requires independence to\n  correlated data without noting it).\n- **Informal arguments** — does the stated reasoning actually support the\n  conclusion drawn, or is there a gap (an unstated assumption doing the\n  real work, a non sequitur, an overgeneralization from one example)?\n\n### 5. Cross-check computed values and figures against the rendered output\n\nDon't trust the source prose's own description of a computed value — verify\nit against what was actually produced:\n\n1. Check whether the project has a **PR-preview deploy**. Look for a sticky\n   preview-link comment on the PR, or check the project's `CLAUDE.md` \u002F\n   `.github\u002Fworkflows\u002F` for a preview workflow (e.g. `preview.yml` +\n   `preview-deploy.yml` built on `rossjrw\u002Fpr-preview-action`, common in this\n   lab's Quarto repos — those deploy to a `gh-pages` branch under\n   `pr-preview\u002Fpr-\u003CN>\u002F\u003Cpath>`).\n2. If a live preview exists, fetch the rendered page (`WebFetch`) and read\n   the actual value\u002Ffigure off it.\n3. If no preview covers the changed page yet, fall back to the project's\n   render command (e.g. `quarto render \u003Cchapter.qmd> --to html` in an R\u002F\n   Quarto repo) and check the local render instead of skipping the check.\n4. Compare what's rendered against what the prose claims. A mismatch is a\n   finding even if the prose \"sounds\" right — the rendered output is ground\n   truth, not the sentence describing it.\n\n### 6. Report\n\nFor every unit checked:\n\n```\n| Claim\u002Fstep | Verdict | Basis (source checked) |\n|------------|---------|-------------------------|\n| \"the estimator is unbiased\" | Inaccurate | Biased under this design per [source\u002Fderivation]; see step 3 below |\n| Derivation step 4 → 5 | Confirmed | Re-derived; algebra holds, dimensions consistent |\n| \"Table 2, row 3 = 4.7\" | Inaccurate | Rendered preview (pr-preview\u002Fpr-N\u002Fchapter.html) shows 4.9 |\n| \"adoption has grown steadily\" | Undefended | No citation or reasoning given; claim itself checks out but needs a source or an explicit basis |\n```\n\nThen, separately, list **additional citations\u002Freferences worth adding** —\nproactively, anywhere a pointer to a source would help a reader, not only\nwhere a claim is currently flagged as uncited.\n\n### 7. Offer to apply corrections\n\nOn request, fix the flagged inaccuracies and reasoning gaps in place with\n`Edit`, preserving the author's intent and voice — same posture as\n`find-ai-tells`'s apply step. Don't silently rewrite without being asked;\nreport first.\n\n## Relationship to other skills\n\n- [`challenge-ambiguous-terminology`](..\u002F..\u002Fshared\u002Fworkflow\u002Fchallenge-ambiguous-terminology.md) —\n  catches phrasing whose meaning is *unresolved*; this skill catches claims\n  and reasoning that are resolved but *wrong*. Run both during review.\n- **`use-preferred-style` \u002F `find-ai-tells`** — style and AI-tell scans; this\n  skill is the accuracy\u002Freasoning counterpart, not a replacement.\n- **`ard` \u002F `ardi`** — when a reviewer (human or bot) flags a factual or\n  reasoning error, `ard` handles the response disposition; this skill is how\n  you (as reviewer) *find* that class of error in the first place.\n- **`prose-fact-checker` agent** — the read-only, parallelizable version of\n  step 2–4 above, for fanning claim-checking out across a `Workflow` when a\n  document has many independent claims to verify at once.\n- **`check-info-quality` (`ciq`)** — this skill verdicts a claim\n  **Accurate \u002F Inaccurate \u002F Undefended**; `ciq`'s check C catches a claim\n  that's accurate by this skill's own standard yet still misleads (missing\n  context, cherry-picking, a citation that only weakly supports it). `ciq`\n  also covers two failure modes this skill doesn't: stale claims (check A)\n  and irrelevant content (check B). Run both.\n\n## Anti-patterns\n\n- **Don't** accept a claim because it sounds authoritative — check it.\n- **Don't** skip a derivation step because \"it's probably fine\" — re-derive\n  it or flag it as unverified, not silently pass it.\n- **Don't** trust the prose's own description of a computed value when a\n  rendered artifact (preview or local render) is available to check it\n  against directly.\n- **Don't** limit citation suggestions to claims already flagged as uncited —\n  suggest them anywhere they'd help a reader, per the fragment's policy.\n",{"name":5,"description":7,"allowed-tools":24,"user-invocable":32},[25,26,27,28,29,30,31],"Bash","Read","Grep","Glob","WebFetch","Edit","Write",true,"main","0f720481f8ce3b3d9fef485f6e230d8b3035748e","skills\u002Ffact-check-prose",1,"https:\u002F\u002Fgithub.com\u002FMorrison-Lab\u002Fai-config","https:\u002F\u002Fmorrison-lab.github.io\u002Fai-config\u002F","https:\u002F\u002Fgithub.com\u002FMorrison-Lab","2026-11-15 04:39:15",{"source":8,"skillName":5,"commands":42,"prompt":52,"zipAvailable":32},[43,46,49],{"tool":44,"command":45},"npx","npx skills add Morrison-Lab\u002Fai-config --skill fact-check-prose",{"tool":47,"command":48},"bunx","bunx skills add Morrison-Lab\u002Fai-config --skill fact-check-prose",{"tool":50,"command":51},"pnpm","pnpm dlx skills add Morrison-Lab\u002Fai-config --skill fact-check-prose","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FMorrison-Lab\u002Fai-config\u002Fblob\u002Fmain\u002Fskills\u002Ffact-check-prose\u002FSKILL.md",[],[55,62,69,76,83,90],{"skillNo":56,"name":57,"nameZh":6,"description":58,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":59,"categories":60,"keywords":61,"publishedAt":19},"01M735BVYCGR5VEC65AFDVSNKW","wrap-up","End-of-session verification and UMS.","https:\u002F\u002Fgithub.com\u002FMorrison-Lab\u002Fai-config\u002Fblob\u002Fmain\u002Fskills\u002Fwrap-up\u002FSKILL.md",[],[],{"skillNo":63,"name":64,"nameZh":6,"description":65,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":66,"categories":67,"keywords":68,"publishedAt":19},"01M735BTH469T1CQ8G2Q9GENCR","workaround-watcher","Watch upstream issue for fix.","https:\u002F\u002Fgithub.com\u002FMorrison-Lab\u002Fai-config\u002Fblob\u002Fmain\u002Fskills\u002Fworkaround-watcher\u002FSKILL.md",[],[],{"skillNo":70,"name":71,"nameZh":6,"description":72,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":73,"categories":74,"keywords":75,"publishedAt":19},"01M735BRKZWFANSQ0GM380B4TD","wn","→ whats-next.","https:\u002F\u002Fgithub.com\u002FMorrison-Lab\u002Fai-config\u002Fblob\u002Fmain\u002Fskills\u002Fwn\u002FSKILL.md",[],[],{"skillNo":77,"name":78,"nameZh":6,"description":79,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":82,"publishedAt":19},"01M735BPE9V830B3B6NGXQ56J6","whats-next","Recommend concrete next step or action.","https:\u002F\u002Fgithub.com\u002FMorrison-Lab\u002Fai-config\u002Fblob\u002Fmain\u002Fskills\u002Fwhats-next\u002FSKILL.md",[],[],{"skillNo":84,"name":85,"nameZh":6,"description":86,"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},"01M735BMHQ60EAD5B7QAWXC5A0","wait-for-results","Wait for pipeline\u002FCI results.","https:\u002F\u002Fgithub.com\u002FMorrison-Lab\u002Fai-config\u002Fblob\u002Fmain\u002Fskills\u002Fwait-for-results\u002FSKILL.md",[],[],{"skillNo":91,"name":92,"nameZh":6,"description":93,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":94,"categories":95,"keywords":96,"publishedAt":19},"01M735BKGZHPN1PEMJ7YBD5VJ8","wai","→ send-notes.","https:\u002F\u002Fgithub.com\u002FMorrison-Lab\u002Fai-config\u002Fblob\u002Fmain\u002Fskills\u002Fwai\u002FSKILL.md",[],[]]