[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6C2CC696DDD3V9AM9D0WVSD":3,"skill-same:01M6C2CC696DDD3V9AM9D0WVSD":68},{"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":25,"publishedAt":40,"linkOnly":41,"markdownContent":42,"markdownBody":43,"frontMatter":44,"refName":47,"commitSha":48,"skillDirectory":49,"forksCount":50,"repositoryHtmlUrl":51,"repositoryHomepageUrl":6,"ownerHtmlUrl":52,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":53,"install":54,"resources":67},"01M6C2CC696DDD3V9AM9D0WVSD","repro",null,"Reproduce a bug, then drive a root-cause-classified fix. Use when asked to \u002Frepro a bug, file a REPRO issue, or escalate a regression. Triggers on the words: repro, reproduce, regression, root cause, \"what broke\", \"file an issue\".","jleechanorg\u002Fjleechan-skills","jleechanorg","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F221404966?v=4","01M6C1VJ99XG950X1DWPEBCBGX","01M6C1VJDK5YVDNT3362RT5WSB",6,"MIT","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Frepro\u002FSKILL.md",[17,22],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000010","后端与接口","SCENARIO","使用场景",{"categoryNo":23,"categoryName":24,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000023","Skill与智能体开发",[5,26,27,28,29,30,31,32,33,34,35,36,37,38,39],"file","regression","reproduce","fix","use","bug","drive","issue","root-cause","triggers","escalate","root","root cause","when","2026-11-10 10:12:30",false,"---\nname: repro\ndescription: \"Reproduce a bug, then drive a root-cause-classified fix. Use when asked to \u002Frepro a bug, file a REPRO issue, or escalate a regression. Triggers on the words: repro, reproduce, regression, root cause, \\\"what broke\\\", \\\"file an issue\\\".\"\nmetadata:\n  type: skill\n---\n\n# \u002Frepro\n\nReproduce a bug end-to-end, classify the root cause, file a REPRO issue, then propose (or drive) the minimal fix. Always root-cause-first; never ship a patch on top of an unverified hypothesis.\n\n## 0. Repo-level skill first — always\n\nBefore anything else, look in the active repository for its own repro workflow: `.claude\u002Fcommands\u002Frepro.md`, `.claude\u002Fskills\u002Frepro*\u002FSKILL.md`, `.claude\u002Fskills\u002Frepro*.md`, and any repro entry in its `AGENTS.md`\u002F`CLAUDE.md`. If one exists, read it and follow it; it wins on every difference within the repository's own scope, subject to the user's explicit instructions and the active permission boundaries, which take precedence over both. This skill applies only where the repository is silent. A repository that routes `\u002Frepro` to its own twin-clone or provider-specific evidence workflow owns that workflow.\n\n## 0.1 Default comparison: original \u002F control \u002F candidate\n\nReport every repro as original (what the real system actually did) \u002F control (replay without the fix) \u002F candidate (replay with the fix) on the same inputs, as defined in `..\u002Frepro-evidence\u002FSKILL.md` § Default comparison, unless the repository skill says otherwise. Reproducing before any fix exists is normal: report original and control, mark the candidate arm `not yet available`, and never invent one.\n\n## Usage\n\n```\n\u002Frepro \u003Cbug description or PR\u002Fissue ref>\n\u002Frepro --issue \u003C#N>           # reproduce from a known REPRO issue\n\u002Frepro --pr \u003C#N>              # reproduce from a known PR (use PR's repro context)\n```\n\n## 1. Understand the report\n\n- Read the bug description, any linked issues\u002FPRs, and any prior `\u002Frepro` outputs.\n- Capture: **symptom** (what the user sees), **reporter** (who\u002Fwhen), **environment** (prod\u002Fdev URL, build SHA, browser\u002Fdevice if known), and **first-seen**.\n- If a regression is suspected, identify the **earliest** merged PR that could plausibly cause it. Use `gh pr list --search \"\u003Ckeyword>\"` and `gh pr view \u003CN> --json files,mergedAt`.\n\n## 2. Try to reproduce (Layer 1 → Layer 4)\n\nClimb the minimal repro ladder. **Stop climbing the ladder only when the blocker is conclusively reproduced.**\n\n| Layer | Tool | When to use |\n|---|---|---|\n| 1. Unit | `pytest \u003Cpath>::\u003Ctest>` (mocked\u002Fisolated) | Cheap; narrows down the unit. Unit-only evidence is **not sufficient** for a production-fix claim. |\n| 2. End-to-end | `pytest \u003Ce2e test>` | Hits real callstack with mocks only at external API boundaries. |\n| 3. MCP \u002F local server | real curl \u002F mcp client against a running local dev server | When the bug needs the real LLM call path. |\n| 4. Browser (final) | `mcp__playwright-mcp` headless against a deployed URL | When the bug is UI\u002Fintegration-specific or only reproduces in a real browser session. |\n\nIf a layer passes but the bug seems environmental, **jump to the highest layer** to verify full stack behavior. Capture exact log lines, screenshots, and HTTP traces.\n\n## 3. Capture smoking gun\n\nMandatory before moving to classification:\n- Exact command(s) run with exit code\n- Key log line(s) — verbatim, with timestamp\n- If UI: screenshot path + the exact element that misbehaves (selector + observed vs. expected)\n- Session id \u002F request id \u002F cdiag event id, if applicable\n\nIf you cannot reproduce, say so explicitly. Do not paper over a non-repro with a \"probably X\" hypothesis.\n\n## 4. File the REPRO issue (if not already filed)\n\nUse this exact title pattern:\n```\nREPRO: \u003Cone-line symptom> — \u003Cone-line root-cause hypothesis>\n```\n\nBody must include:\n- **Summary** — one paragraph\n- **Smoking gun** — verbatim log\u002Fscreenshot\u002FHTTP trace\n- **Root cause candidates** — ranked by likelihood (see §7)\n- **How to find the real cause** — what extra instrumentation would disambiguate\n- **Steps to reproduce** — exact URLs \u002F commands\n- **Related tracking** — PRs, issues, prior fix attempts (with their SHA \u002F merge commit)\n\n## 5. Drive the fix (only after root cause is confirmed)\n\n1. Update the issue's \"Root cause candidates\" → \"Confirmed root cause\" section with the evidence.\n2. Branch from `origin\u002Fmain` (`git fetch origin && git checkout -B \u003Cbranch> origin\u002Fmain`).\n3. **Fix prompt\u002Fschema\u002Finstruction first** before adding backend protection, fallback, clamp, sanitizer, or retry. Backend protection only after documenting why prompt correction is insufficient.\n4. Add tests that would have caught the bug (Layer 1 unit + Layer 2 e2e at minimum).\n5. Open PR with a body that links the REPRO issue, names the root cause, and shows the new test failing on `origin\u002Fmain` (RED) and passing on the branch (GREEN).\n6. Drive the PR to green; do not stop at \"I made the change locally\".\n\n## 6. Honest reporting\n\n- Distinguish **observed** from **recommended** at all times.\n- If a fix is blocked, say what blocks it (auth, missing evidence, conflicting requirement) — not a vague \"I will follow up\".\n- Past fix attempts that were reverted must be cited: their SHA, the analysis that was wrong, and what the revert taught us.\n\n## 7. Root-cause classification (NEW — 2026-06-16)\n\nBefore proposing a fix, **classify the root cause into exactly one of these buckets**. Pick the most specific bucket that fits; the bucket drives which fix-pattern to apply.\n\n### 7.1 Prompt \u002F instruction\n- The LLM was given an ambiguous, contradictory, or underspecified prompt.\n- Same call with a clearer prompt succeeds.\n- **Fix pattern:** clarify the prompt \u002F schema \u002F instruction first. Do not add backend defense.\n\n### 7.2 Backend \u002F business logic\n- The server-side code is doing the wrong thing deterministically (bug, race, stale data, missing migration).\n- Repro is stable across prompt variations.\n- **Fix pattern:** patch the backend. Add a unit test that exercises the code path with the exact input shape from the smoking gun.\n\n### 7.3 External API \u002F provider\n- The LLM provider or third-party API returned a wrong\u002Fempty\u002Fmalformed response.\n- The shape is \"we sent X, we got Y, and the same call 5 minutes later succeeds\".\n- **Fix pattern:** retry with backoff (only if the call is idempotent) + log the provider's response. Do not paper over with a clamp\u002Fsanitizer unless we can prove the response is unrecoverable.\n\n### 7.4 UI \u002F integration\n- The bug only reproduces in a real browser session, and the server logs show no anomaly.\n- The same data flow works in unit + e2e + mcp layers.\n- **Fix pattern:** fix the UI \u002F event handler \u002F state machine. Add a Playwright test that loads the real URL and asserts the expected DOM state.\n\n### 7.5 Environment \u002F config\n- A config value, env var, or deployment setting is wrong, missing, or stale.\n- Switching environments (dev → prod) or fixing the config eliminates the bug.\n- **Fix pattern:** fix the config, audit all environments, add a startup assertion that fails loud if the config is missing.\n\n### 7.6 Unknown \u002F under-instrumented\n- We have a symptom but no smoking gun. Logs do not disambiguate.\n- **Fix pattern (do NOT propose a fix yet):** add the missing instrumentation first. Ship a no-op PR that adds the events \u002F logs that would have caught this. Then re-`\u002Frepro` with the new evidence.\n\n### 7.7 Classification rules (enforced)\n\n- **Root-cause-first invocation:** If classification is **unknown \u002F under-instrumented**, the next action is **always** \"add instrumentation\", not \"guess a fix\". Guessing a fix from a weak signal is the exact anti-pattern that produced prior reverted fixes (e.g. PR #7551's 250ms grace, which addressed the wrong race).\n- **One root cause per bug:** Pick exactly one bucket. If you think it's both prompt AND backend, you have not classified yet — go back to §2 and reproduce.\n- **Past reverted fixes are evidence:** Before proposing a fix, search the repo for prior fix attempts of the same symptom. If any were reverted, cite them in the REPRO issue and explain why this attempt is different.\n\n## 8. Output template (always end with this)\n\n```\n## REPRO summary\n- Symptom:\n- Root cause (bucket 7.x):\n- Smoking gun (file:line \u002F log \u002F cdiag id):\n- Fix pattern: [prompt | backend | retry+sanitize | UI | config | instrumentation-first]\n- Status: [REPRODUCED | NOT_REPRODUCED | UNDER_INSTRUMENTED]\n- Next action: [link to issue \u002F PR \u002F next skill]\n```\n","# \u002Frepro\n\nReproduce a bug end-to-end, classify the root cause, file a REPRO issue, then propose (or drive) the minimal fix. Always root-cause-first; never ship a patch on top of an unverified hypothesis.\n\n## 0. Repo-level skill first — always\n\nBefore anything else, look in the active repository for its own repro workflow: `.claude\u002Fcommands\u002Frepro.md`, `.claude\u002Fskills\u002Frepro*\u002FSKILL.md`, `.claude\u002Fskills\u002Frepro*.md`, and any repro entry in its `AGENTS.md`\u002F`CLAUDE.md`. If one exists, read it and follow it; it wins on every difference within the repository's own scope, subject to the user's explicit instructions and the active permission boundaries, which take precedence over both. This skill applies only where the repository is silent. A repository that routes `\u002Frepro` to its own twin-clone or provider-specific evidence workflow owns that workflow.\n\n## 0.1 Default comparison: original \u002F control \u002F candidate\n\nReport every repro as original (what the real system actually did) \u002F control (replay without the fix) \u002F candidate (replay with the fix) on the same inputs, as defined in `..\u002Frepro-evidence\u002FSKILL.md` § Default comparison, unless the repository skill says otherwise. Reproducing before any fix exists is normal: report original and control, mark the candidate arm `not yet available`, and never invent one.\n\n## Usage\n\n```\n\u002Frepro \u003Cbug description or PR\u002Fissue ref>\n\u002Frepro --issue \u003C#N>           # reproduce from a known REPRO issue\n\u002Frepro --pr \u003C#N>              # reproduce from a known PR (use PR's repro context)\n```\n\n## 1. Understand the report\n\n- Read the bug description, any linked issues\u002FPRs, and any prior `\u002Frepro` outputs.\n- Capture: **symptom** (what the user sees), **reporter** (who\u002Fwhen), **environment** (prod\u002Fdev URL, build SHA, browser\u002Fdevice if known), and **first-seen**.\n- If a regression is suspected, identify the **earliest** merged PR that could plausibly cause it. Use `gh pr list --search \"\u003Ckeyword>\"` and `gh pr view \u003CN> --json files,mergedAt`.\n\n## 2. Try to reproduce (Layer 1 → Layer 4)\n\nClimb the minimal repro ladder. **Stop climbing the ladder only when the blocker is conclusively reproduced.**\n\n| Layer | Tool | When to use |\n|---|---|---|\n| 1. Unit | `pytest \u003Cpath>::\u003Ctest>` (mocked\u002Fisolated) | Cheap; narrows down the unit. Unit-only evidence is **not sufficient** for a production-fix claim. |\n| 2. End-to-end | `pytest \u003Ce2e test>` | Hits real callstack with mocks only at external API boundaries. |\n| 3. MCP \u002F local server | real curl \u002F mcp client against a running local dev server | When the bug needs the real LLM call path. |\n| 4. Browser (final) | `mcp__playwright-mcp` headless against a deployed URL | When the bug is UI\u002Fintegration-specific or only reproduces in a real browser session. |\n\nIf a layer passes but the bug seems environmental, **jump to the highest layer** to verify full stack behavior. Capture exact log lines, screenshots, and HTTP traces.\n\n## 3. Capture smoking gun\n\nMandatory before moving to classification:\n- Exact command(s) run with exit code\n- Key log line(s) — verbatim, with timestamp\n- If UI: screenshot path + the exact element that misbehaves (selector + observed vs. expected)\n- Session id \u002F request id \u002F cdiag event id, if applicable\n\nIf you cannot reproduce, say so explicitly. Do not paper over a non-repro with a \"probably X\" hypothesis.\n\n## 4. File the REPRO issue (if not already filed)\n\nUse this exact title pattern:\n```\nREPRO: \u003Cone-line symptom> — \u003Cone-line root-cause hypothesis>\n```\n\nBody must include:\n- **Summary** — one paragraph\n- **Smoking gun** — verbatim log\u002Fscreenshot\u002FHTTP trace\n- **Root cause candidates** — ranked by likelihood (see §7)\n- **How to find the real cause** — what extra instrumentation would disambiguate\n- **Steps to reproduce** — exact URLs \u002F commands\n- **Related tracking** — PRs, issues, prior fix attempts (with their SHA \u002F merge commit)\n\n## 5. Drive the fix (only after root cause is confirmed)\n\n1. Update the issue's \"Root cause candidates\" → \"Confirmed root cause\" section with the evidence.\n2. Branch from `origin\u002Fmain` (`git fetch origin && git checkout -B \u003Cbranch> origin\u002Fmain`).\n3. **Fix prompt\u002Fschema\u002Finstruction first** before adding backend protection, fallback, clamp, sanitizer, or retry. Backend protection only after documenting why prompt correction is insufficient.\n4. Add tests that would have caught the bug (Layer 1 unit + Layer 2 e2e at minimum).\n5. Open PR with a body that links the REPRO issue, names the root cause, and shows the new test failing on `origin\u002Fmain` (RED) and passing on the branch (GREEN).\n6. Drive the PR to green; do not stop at \"I made the change locally\".\n\n## 6. Honest reporting\n\n- Distinguish **observed** from **recommended** at all times.\n- If a fix is blocked, say what blocks it (auth, missing evidence, conflicting requirement) — not a vague \"I will follow up\".\n- Past fix attempts that were reverted must be cited: their SHA, the analysis that was wrong, and what the revert taught us.\n\n## 7. Root-cause classification (NEW — 2026-06-16)\n\nBefore proposing a fix, **classify the root cause into exactly one of these buckets**. Pick the most specific bucket that fits; the bucket drives which fix-pattern to apply.\n\n### 7.1 Prompt \u002F instruction\n- The LLM was given an ambiguous, contradictory, or underspecified prompt.\n- Same call with a clearer prompt succeeds.\n- **Fix pattern:** clarify the prompt \u002F schema \u002F instruction first. Do not add backend defense.\n\n### 7.2 Backend \u002F business logic\n- The server-side code is doing the wrong thing deterministically (bug, race, stale data, missing migration).\n- Repro is stable across prompt variations.\n- **Fix pattern:** patch the backend. Add a unit test that exercises the code path with the exact input shape from the smoking gun.\n\n### 7.3 External API \u002F provider\n- The LLM provider or third-party API returned a wrong\u002Fempty\u002Fmalformed response.\n- The shape is \"we sent X, we got Y, and the same call 5 minutes later succeeds\".\n- **Fix pattern:** retry with backoff (only if the call is idempotent) + log the provider's response. Do not paper over with a clamp\u002Fsanitizer unless we can prove the response is unrecoverable.\n\n### 7.4 UI \u002F integration\n- The bug only reproduces in a real browser session, and the server logs show no anomaly.\n- The same data flow works in unit + e2e + mcp layers.\n- **Fix pattern:** fix the UI \u002F event handler \u002F state machine. Add a Playwright test that loads the real URL and asserts the expected DOM state.\n\n### 7.5 Environment \u002F config\n- A config value, env var, or deployment setting is wrong, missing, or stale.\n- Switching environments (dev → prod) or fixing the config eliminates the bug.\n- **Fix pattern:** fix the config, audit all environments, add a startup assertion that fails loud if the config is missing.\n\n### 7.6 Unknown \u002F under-instrumented\n- We have a symptom but no smoking gun. Logs do not disambiguate.\n- **Fix pattern (do NOT propose a fix yet):** add the missing instrumentation first. Ship a no-op PR that adds the events \u002F logs that would have caught this. Then re-`\u002Frepro` with the new evidence.\n\n### 7.7 Classification rules (enforced)\n\n- **Root-cause-first invocation:** If classification is **unknown \u002F under-instrumented**, the next action is **always** \"add instrumentation\", not \"guess a fix\". Guessing a fix from a weak signal is the exact anti-pattern that produced prior reverted fixes (e.g. PR #7551's 250ms grace, which addressed the wrong race).\n- **One root cause per bug:** Pick exactly one bucket. If you think it's both prompt AND backend, you have not classified yet — go back to §2 and reproduce.\n- **Past reverted fixes are evidence:** Before proposing a fix, search the repo for prior fix attempts of the same symptom. If any were reverted, cite them in the REPRO issue and explain why this attempt is different.\n\n## 8. Output template (always end with this)\n\n```\n## REPRO summary\n- Symptom:\n- Root cause (bucket 7.x):\n- Smoking gun (file:line \u002F log \u002F cdiag id):\n- Fix pattern: [prompt | backend | retry+sanitize | UI | config | instrumentation-first]\n- Status: [REPRODUCED | NOT_REPRODUCED | UNDER_INSTRUMENTED]\n- Next action: [link to issue \u002F PR \u002F next skill]\n```\n",{"name":5,"metadata":45,"description":7},{"type":46},"skill","main","395c0f62cafdf72b2941a9aeb90a63ac6d047925",".claude\u002Fskills\u002Frepro",0,"https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills","https:\u002F\u002Fgithub.com\u002Fjleechanorg","2026-11-12 14:45:07",{"source":8,"skillName":5,"commands":55,"prompt":65,"zipAvailable":66},[56,59,62],{"tool":57,"command":58},"npx","npx skills add jleechanorg\u002Fjleechan-skills --skill repro",{"tool":60,"command":61},"bunx","bunx skills add jleechanorg\u002Fjleechan-skills --skill repro",{"tool":63,"command":64},"pnpm","pnpm dlx skills add jleechanorg\u002Fjleechan-skills --skill repro","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Frepro\u002FSKILL.md",true,[],[69,85,107,121,135,151],{"skillNo":70,"name":71,"nameZh":6,"description":72,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":73,"categories":74,"keywords":75,"publishedAt":40},"01M716Y5MTJ6GYJZ2ATXCJK8G4","instinct","Use when the user invokes \u002Finstinct or \u002Fin, or asks to \"message Instinct\", \"ask Instinct\", \"tell Instinct\", \"check Instinct\", or read replies from the configured peer assistant. iMessage first, WhatsApp Web as fallback.","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Finstinct\u002FSKILL.md",[],[71,76,39,77,78,30,79,80,81,82,83,84],"user","replies","check","identity","whatsapp","gmail","read","message","ask",{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":96,"publishedAt":40},"01M716Y4HF7FVVMJSA2CCVN5VF","hr","Compatibility alias for history-resume; produce the same fresh-state, provenance-aware resume briefing.","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhr\u002FSKILL.md",[91],{"categoryNo":92,"categoryName":93,"taxonomyType":94,"taxonomyTypeDesc":95},"01JBSEEDCAT000000000000038","人力资源","PROFESSION","适用职业",[87,97,98,99,100,101,102,103,104,105,106],"resume","compatibility","provenance","for","history","produce","alias","briefing","state","fresh",{"skillNo":108,"name":109,"nameZh":6,"description":110,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":111,"categories":112,"keywords":114,"publishedAt":40},"01M716Y1FFNFKFJYYKF8HQA6T8","history-resume","Resume briefing for the current repository branch or worktree using fresh state and admitted history sources.","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhistory-resume\u002FSKILL.md",[113],{"categoryNo":92,"categoryName":93,"taxonomyType":94,"taxonomyTypeDesc":95},[101,97,115,116,76,117,104,118,100,119,120,105,106],"history resume","branch","analysis","repository","worktree","sources",{"skillNo":122,"name":123,"nameZh":6,"description":124,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":125,"categories":126,"keywords":127,"publishedAt":40},"01M716XWXJ8EPHK4RH9T14TR85","grok","Alias for the canonical grokbot browser conversation workflow; \u002Fgrok talks to the Grok bot the same way \u002Fdot talks to Dot.","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fgrok\u002FSKILL.md",[],[123,128,129,130,100,131,103,132,133,134],"dot","bot","workflow","canonical","conversation","browser","grokbot",{"skillNo":136,"name":137,"nameZh":6,"description":138,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":139,"categories":140,"keywords":141,"publishedAt":40},"01M70N9Q07DTE0EW5KFXMMMSJ0","codexa-bench","Compare `codexa` and normal `codex` on a fair split of comparable work by default; use `--strict` for duplicate runs of the exact same task.","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fcodexa-bench\u002FSKILL.md",[],[142,143,144,30,145,100,146,147,148,149,150],"codexa bench","work","split","default","duplicate","task","compare","normal","codex",{"skillNo":152,"name":153,"nameZh":6,"description":154,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":155,"licenseFileUrl":156,"sourceHtmlUrl":157,"categories":158,"keywords":168,"publishedAt":40},"01M6C24S2X60SP8BBFGAEKPEGW","frontend-design","Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, artifacts, posters, or applications (examples include websites, landing pages, dashboards, React components, HTML\u002FCSS layouts, or when styling\u002Fbeautifying any web UI). Generates creative, polished code and UI design that avoids generic AI aesthetics.","Apache-2.0","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Ffrontend-design\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Ffrontend-design\u002FSKILL.md",[159,162,165],{"categoryNo":160,"categoryName":161,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000009","前端开发",{"categoryNo":163,"categoryName":164,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000015","视觉设计",{"categoryNo":166,"categoryName":167,"taxonomyType":94,"taxonomyTypeDesc":95},"01JBSEEDCAT000000000000032","设计师",[169,170,171,30,172,173,174,76,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,39,193],"frontend","design","frontend design","generic","ui design","react","html","components","code","artifacts","ui","aesthetics","Production","web components","landing","applications","quality","web","css","create","examples","with","react components","web ui","pages"]