[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M77HY54D6CN18VRZ95WKP0AE":3,"skill-same:01M77HY54D6CN18VRZ95WKP0AE":70},{"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":24,"commitSha":25,"skillDirectory":26,"forksCount":13,"repositoryHtmlUrl":27,"repositoryHomepageUrl":6,"ownerHtmlUrl":28,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":29,"install":30,"resources":43},"01M77HY54D6CN18VRZ95WKP0AE","code-review",null,"Review the current diff, or a PR number\u002Fbranch\u002Fpath target, for correctness bugs (plus reuse\u002Fsimplification\u002Fefficiency cleanups where the model's recipe covers them) at the effort level given first, before any target (low|medium|high|xhigh|max: from few high-confidence findings to many, some uncertain; ultra: deep multi-agent review in the cloud (requires claude.ai account access)); with no level given, it reuses the level you typed last. Pass --comment to post findings as inline PR comments, or --fix to apply them to the working tree after the review. Pass --max-findings \u003Cn> to report up to n findings, or --max-findings all for every finding. The choice stays until you pass --max-findings default. For ultra on a GitHub.com PR target, --post asks to post the finished review's findings to the PR as a single comment from the user's GitHub account (not a review; the launch dialog still confirms in interactive sessions, while non-interactive mode posts on the flag alone) and --no-post hides that option.","raaOS\u002Fsystem_prompts_leaks","raaOS","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F214067397?v=4","01M77HV6K4ZTA4BZAGX00GTW6J","01M77HV6KX55HVRY679V25JXF8",0,"CC0-1.0","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FAnthropic\u002Fclaude-code\u002Fskills\u002Fcode-review\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: code-review\ndescription: |-\n  Review the current diff, or a PR number\u002Fbranch\u002Fpath target, for correctness bugs (plus reuse\u002Fsimplification\u002Fefficiency cleanups where the model's recipe covers them) at the effort level given first, before any target (low|medium|high|xhigh|max: from few high-confidence findings to many, some uncertain; ultra: deep multi-agent review in the cloud (requires claude.ai account access)); with no level given, it reuses the level you typed last. Pass --comment to post findings as inline PR comments, or --fix to apply them to the working tree after the review. Pass --max-findings \u003Cn> to report up to n findings, or --max-findings all for every finding. The choice stays until you pass --max-findings default. For ultra on a GitHub.com PR target, --post asks to post the finished review's findings to the PR as a single comment from the user's GitHub account (not a review; the launch dialog still confirms in interactive sessions, while non-interactive mode posts on the flag alone) and --no-post hides that option.\nwhen_to_use: |-\n  Use the max or ultra level only when the user, or instructions such as CLAUDE.md or a skill, explicitly ask for one, never on your own judgment: they are the costliest reviews and start many reviewer subagents.\n---\n\n`high effort → 8 inline angles → dedup (no verify) → ≤10 findings`\n\nYou are reviewing for **recall** at high effort: catch every real bug a careful\nreviewer would catch in one sitting. At this level, catching real bugs matters\nmore than avoiding false positives. Err on the side of surfacing.\n\n## Phase 0 — Gather the diff\n\nRun `git diff @{upstream}...HEAD` (or `git diff main...HEAD` \u002F `git diff HEAD~1`\nif there's no upstream) to get the unified diff under review. If there are\nuncommitted changes, or the range diff is empty, also run `git diff HEAD` and\ninclude the working-tree changes in scope — the review often runs before the\ncommit. If a PR number, branch name, or file path was passed as an argument,\nreview that target instead. Treat this diff as the review scope.\n\n## Phase 1 — Find candidates (3 correctness angles + 3 cleanup angles + 1 altitude angle + 1 conventions angle, up to 6 each)\n\nRun **8 independent finder angles** in sequence yourself, in THIS context — do NOT spawn subagents for them. Each\nsurfaces **up to 6 candidate findings** with `file`, `line`, a one-line\n`summary`, and a concrete `failure_scenario`.\n\n### Angle A — line-by-line diff scan\n\nRead every hunk in the diff, line by line. Then Read the enclosing function for\neach hunk — bugs in unchanged lines of a touched function are in scope (the PR\nre-exposes or fails to fix them). For every line ask: what input, state, timing,\nor platform makes this line wrong? Look for inverted\u002Fwrong conditions,\noff-by-one, null\u002Fundefined deref, missing `await`, falsy-zero checks,\nwrong-variable copy-paste, error swallowed in catch, unescaped regex metachars.\n\n### Angle B — removed-behavior auditor\n\nFor every line the diff DELETES or replaces, name the invariant or behavior it\nenforced, then search the new code for where that invariant is re-established.\nIf you can't find it, that's a candidate: a removed guard, a dropped error\npath, a narrowed validation, a deleted test that was covering a real case.\n\n### Angle C — cross-file tracer\n\nFor each function the diff changes, find its callers (Grep for the symbol) and\ncheck whether the change breaks any call site: a new precondition, a changed\nreturn shape, a new exception, a timing\u002Fordering dependency. Also check callees:\ndoes a parallel change in the same PR make a call unsafe?\n\n### Reuse\n\nThe angles above hunt for bugs; this one and the next two hunt for cleanup in\nthe changed code. Flag new code that re-implements something the codebase\nalready has — Grep shared\u002Futility modules and files adjacent to the change,\nand name the existing helper to call instead.\n\n### Simplification\n\nFlag unnecessary complexity the diff adds: redundant or derivable state,\ncopy-paste with slight variation, deep nesting, dead code left behind. Name\nthe simpler form that does the same job.\n\n### Efficiency\n\nFlag wasted work the diff introduces: redundant computation or repeated I\u002FO,\nindependent operations run sequentially, blocking work added to startup or\nhot paths. Also flag long-lived objects built from closures or captured\nenvironments — they keep the entire enclosing scope alive for the object's\nlifetime (a memory leak when that scope holds large values); prefer a\nclass\u002Fstruct that copies only the fields it needs. Name the cheaper\nalternative.\n\n### Altitude\n\nCheck that each change fixes the root cause at the right depth rather than\npatching a symptom with a fragile bandaid. Special cases layered on shared\ninfrastructure are a sign the fix isn't deep enough — prefer the simpler, more\ngeneral change to the underlying mechanism over adding special cases, and name\nthat change.\n\n### Conventions (CLAUDE.md)\n\nFind the CLAUDE.md files that govern the changed code: the user-level\n~\u002F.claude\u002FCLAUDE.md, the repo-root CLAUDE.md, plus any CLAUDE.md or\nCLAUDE.local.md in a directory that is an ancestor of a changed file (a\ndirectory's CLAUDE.md only applies to files at or below it). Read each one\nthat exists, then check the diff for clear violations of the rules they state.\n\nOnly flag a violation when you can quote the exact rule and the exact line\nthat breaks it — no style preferences, no vague \"spirit of the doc\"\ninferences. In the finding, name the CLAUDE.md path and quote the rule so the\nreport can cite it. If no CLAUDE.md applies, return nothing for this angle.\n\nCleanup, altitude, and conventions candidates use the same\n`file`\u002F`line`\u002F`summary` shape; in `failure_scenario`, state the concrete\ncost (what is duplicated, wasted, harder to maintain, or which CLAUDE.md rule\nis broken) instead of a crash. Correctness bugs always outrank cleanup,\naltitude, and conventions findings when the output cap forces a cut.\n\nPass every candidate with a nameable failure scenario through — finders that\nsilently drop half-believed candidates are the dominant cause of misses.\n\n## Phase 2 — Dedup only (no verify)\n\nPool all candidates. Dedup near-duplicates only (same defect, same location, same reason → keep one). Do NOT run verifiers; do NOT re-judge. Sort by severity.\n\n## Output\n\nTarget **at least 5 findings**. If fewer genuine findings exist, emit what you have — do not invent to hit the floor.\n\nReturn findings as a JSON array of at most 10 objects:\n\n```json\n[\n  {\n    \"file\": \"path\u002Fto\u002Ffile.ext\",\n    \"line\": 123,\n    \"summary\": \"one-sentence statement of the bug\",\n    \"failure_scenario\": \"concrete inputs\u002Fstate → wrong output\u002Fcrash\"\n  }\n]\n```\n\nRanked most-severe first. If more than 10 survive, keep the 10 most\nsevere. If nothing survives, return `[]`. Do not call the\nReportFindings tool even if it is available - this review's\noutput contract is the JSON block above.\n","`high effort → 8 inline angles → dedup (no verify) → ≤10 findings`\n\nYou are reviewing for **recall** at high effort: catch every real bug a careful\nreviewer would catch in one sitting. At this level, catching real bugs matters\nmore than avoiding false positives. Err on the side of surfacing.\n\n## Phase 0 — Gather the diff\n\nRun `git diff @{upstream}...HEAD` (or `git diff main...HEAD` \u002F `git diff HEAD~1`\nif there's no upstream) to get the unified diff under review. If there are\nuncommitted changes, or the range diff is empty, also run `git diff HEAD` and\ninclude the working-tree changes in scope — the review often runs before the\ncommit. If a PR number, branch name, or file path was passed as an argument,\nreview that target instead. Treat this diff as the review scope.\n\n## Phase 1 — Find candidates (3 correctness angles + 3 cleanup angles + 1 altitude angle + 1 conventions angle, up to 6 each)\n\nRun **8 independent finder angles** in sequence yourself, in THIS context — do NOT spawn subagents for them. Each\nsurfaces **up to 6 candidate findings** with `file`, `line`, a one-line\n`summary`, and a concrete `failure_scenario`.\n\n### Angle A — line-by-line diff scan\n\nRead every hunk in the diff, line by line. Then Read the enclosing function for\neach hunk — bugs in unchanged lines of a touched function are in scope (the PR\nre-exposes or fails to fix them). For every line ask: what input, state, timing,\nor platform makes this line wrong? Look for inverted\u002Fwrong conditions,\noff-by-one, null\u002Fundefined deref, missing `await`, falsy-zero checks,\nwrong-variable copy-paste, error swallowed in catch, unescaped regex metachars.\n\n### Angle B — removed-behavior auditor\n\nFor every line the diff DELETES or replaces, name the invariant or behavior it\nenforced, then search the new code for where that invariant is re-established.\nIf you can't find it, that's a candidate: a removed guard, a dropped error\npath, a narrowed validation, a deleted test that was covering a real case.\n\n### Angle C — cross-file tracer\n\nFor each function the diff changes, find its callers (Grep for the symbol) and\ncheck whether the change breaks any call site: a new precondition, a changed\nreturn shape, a new exception, a timing\u002Fordering dependency. Also check callees:\ndoes a parallel change in the same PR make a call unsafe?\n\n### Reuse\n\nThe angles above hunt for bugs; this one and the next two hunt for cleanup in\nthe changed code. Flag new code that re-implements something the codebase\nalready has — Grep shared\u002Futility modules and files adjacent to the change,\nand name the existing helper to call instead.\n\n### Simplification\n\nFlag unnecessary complexity the diff adds: redundant or derivable state,\ncopy-paste with slight variation, deep nesting, dead code left behind. Name\nthe simpler form that does the same job.\n\n### Efficiency\n\nFlag wasted work the diff introduces: redundant computation or repeated I\u002FO,\nindependent operations run sequentially, blocking work added to startup or\nhot paths. Also flag long-lived objects built from closures or captured\nenvironments — they keep the entire enclosing scope alive for the object's\nlifetime (a memory leak when that scope holds large values); prefer a\nclass\u002Fstruct that copies only the fields it needs. Name the cheaper\nalternative.\n\n### Altitude\n\nCheck that each change fixes the root cause at the right depth rather than\npatching a symptom with a fragile bandaid. Special cases layered on shared\ninfrastructure are a sign the fix isn't deep enough — prefer the simpler, more\ngeneral change to the underlying mechanism over adding special cases, and name\nthat change.\n\n### Conventions (CLAUDE.md)\n\nFind the CLAUDE.md files that govern the changed code: the user-level\n~\u002F.claude\u002FCLAUDE.md, the repo-root CLAUDE.md, plus any CLAUDE.md or\nCLAUDE.local.md in a directory that is an ancestor of a changed file (a\ndirectory's CLAUDE.md only applies to files at or below it). Read each one\nthat exists, then check the diff for clear violations of the rules they state.\n\nOnly flag a violation when you can quote the exact rule and the exact line\nthat breaks it — no style preferences, no vague \"spirit of the doc\"\ninferences. In the finding, name the CLAUDE.md path and quote the rule so the\nreport can cite it. If no CLAUDE.md applies, return nothing for this angle.\n\nCleanup, altitude, and conventions candidates use the same\n`file`\u002F`line`\u002F`summary` shape; in `failure_scenario`, state the concrete\ncost (what is duplicated, wasted, harder to maintain, or which CLAUDE.md rule\nis broken) instead of a crash. Correctness bugs always outrank cleanup,\naltitude, and conventions findings when the output cap forces a cut.\n\nPass every candidate with a nameable failure scenario through — finders that\nsilently drop half-believed candidates are the dominant cause of misses.\n\n## Phase 2 — Dedup only (no verify)\n\nPool all candidates. Dedup near-duplicates only (same defect, same location, same reason → keep one). Do NOT run verifiers; do NOT re-judge. Sort by severity.\n\n## Output\n\nTarget **at least 5 findings**. If fewer genuine findings exist, emit what you have — do not invent to hit the floor.\n\nReturn findings as a JSON array of at most 10 objects:\n\n```json\n[\n  {\n    \"file\": \"path\u002Fto\u002Ffile.ext\",\n    \"line\": 123,\n    \"summary\": \"one-sentence statement of the bug\",\n    \"failure_scenario\": \"concrete inputs\u002Fstate → wrong output\u002Fcrash\"\n  }\n]\n```\n\nRanked most-severe first. If more than 10 survive, keep the 10 most\nsevere. If nothing survives, return `[]`. Do not call the\nReportFindings tool even if it is available - this review's\noutput contract is the JSON block above.\n",{"name":5,"description":7,"when_to_use":23},"Use the max or ultra level only when the user, or instructions such as CLAUDE.md or a skill, explicitly ask for one, never on your own judgment: they are the costliest reviews and start many reviewer subagents.","main","2975a5ac469614541ae22929d43b6da2635e1ee9","Anthropic\u002Fclaude-code\u002Fskills\u002Fcode-review","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks","https:\u002F\u002Fgithub.com\u002FraaOS","2026-11-14 21:19:17",{"source":8,"skillName":5,"commands":31,"prompt":41,"zipAvailable":42},[32,35,38],{"tool":33,"command":34},"npx","npx skills add raaOS\u002Fsystem_prompts_leaks --skill code-review",{"tool":36,"command":37},"bunx","bunx skills add raaOS\u002Fsystem_prompts_leaks --skill code-review",{"tool":39,"command":40},"pnpm","pnpm dlx skills add raaOS\u002Fsystem_prompts_leaks --skill code-review","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FAnthropic\u002Fclaude-code\u002Fskills\u002Fcode-review\u002FSKILL.md",true,[44,50,54,58,62,66],{"path":45,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":48,"sourceHtmlUrl":49},"high.md","REFERENCE","参考文档",5658,"https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FAnthropic\u002Fclaude-code\u002Fskills\u002Fcode-review\u002Fhigh.md",{"path":51,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":52,"sourceHtmlUrl":53},"low.md",1271,"https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FAnthropic\u002Fclaude-code\u002Fskills\u002Fcode-review\u002Flow.md",{"path":55,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":56,"sourceHtmlUrl":57},"max.md",7713,"https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FAnthropic\u002Fclaude-code\u002Fskills\u002Fcode-review\u002Fmax.md",{"path":59,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":60,"sourceHtmlUrl":61},"medium.md",1590,"https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FAnthropic\u002Fclaude-code\u002Fskills\u002Fcode-review\u002Fmedium.md",{"path":63,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":64,"sourceHtmlUrl":65},"report-findings-tool.md",3031,"https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FAnthropic\u002Fclaude-code\u002Fskills\u002Fcode-review\u002Freport-findings-tool.md",{"path":67,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":68,"sourceHtmlUrl":69},"xhigh.md",7163,"https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FAnthropic\u002Fclaude-code\u002Fskills\u002Fcode-review\u002Fxhigh.md",[71,77,84,91,98,105],{"skillNo":72,"name":73,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":18},"01M77JV1M0ZZTZZREM9Q96TKZZ","model-council","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FPerplexity\u002Fskills\u002Fmodel-council\u002FSKILL.md",[],[],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":18},"01M77JTYND3ETS9EGCX3B9CW6E","writing-style","Draft or edit writing in the user's voice for the person who will read it and where it will appear. Use for messages, emails, documents or slides written on their behalf, not for your own replies.","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FOpenAI\u002Fdots\u002Fskills\u002Fwriting-style\u002FSKILL.md",[],[],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":18},"01M77JTVWEHY80JC2H1JG1JMQB","write-page","Create or edit requested Page\u002FSpace content, or use for prose you have already decided to save as a standalone Markdown file, only if the user did not explicitly request a Markdown file. Honor established formats, destinations, and repository documentation. A selected Page alone does not authorize a write.","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FOpenAI\u002Fdots\u002Fskills\u002Fwrite-page\u002FSKILL.md",[],[],{"skillNo":92,"name":93,"nameZh":6,"description":94,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":95,"categories":96,"keywords":97,"publishedAt":18},"01M77JTQ5ND23QPF6RJYJPXHSZ","software-engineering","For use only in threads for this proactive personal assistant. Investigate software issues; inspect, write, review, or test code; work on a repository or local engineering file; or create, fix, or monitor a pull request or CI run. Use for actual engineering or product QA work, not a general programming explanation.","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FOpenAI\u002Fdots\u002Fskills\u002Fsoftware-engineering\u002FSKILL.md",[],[],{"skillNo":99,"name":100,"nameZh":6,"description":101,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":102,"categories":103,"keywords":104,"publishedAt":18},"01M77JTM3ZDPCYE8C1FW7G7S2B","slides-artifact","Use when the user asks to create, edit, or review slides or a presentation (PowerPoint or Google Slides), including when the task clearly calls for an audience-ready slide deck. Also use to answer questions that require inspecting an existing deck; not for general presentation advice or a chat-only outline.","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FOpenAI\u002Fdots\u002Fskills\u002Fslides-artifact\u002FSKILL.md",[],[],{"skillNo":106,"name":107,"nameZh":6,"description":108,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":109,"categories":110,"keywords":111,"publishedAt":18},"01M77JTHTWZN2J27KD2H66XDBZ","slack","Use when reading Slack conversations, deciding when to respond, or sending messages, files, and reactions in Slack.","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FOpenAI\u002Fdots\u002Fskills\u002Fslack\u002FSKILL.md",[],[]]