[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6BZENS3W728KRG2F3EX3ZD3":3,"skill-same:01M6BZENS3W728KRG2F3EX3ZD3":65},{"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":37,"linkOnly":38,"markdownContent":39,"markdownBody":40,"frontMatter":41,"refName":42,"commitSha":43,"skillDirectory":44,"forksCount":45,"repositoryHtmlUrl":46,"repositoryHomepageUrl":47,"ownerHtmlUrl":48,"ownerBlogUrl":49,"ownerDescription":6,"updatedAt":50,"install":51,"resources":64},"01M6BZENS3W728KRG2F3EX3ZD3","codex",null,"Read before running the codex CLI (GPT-6 astra) — the additional reviewer of the pr skill's self-review, plan attacks, debugging, scoped implementation. Only when codex is logged in and has quota.","clice-io\u002Fclice","clice-io","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F176553789?v=4","01M6BZEJH3SKDZ3SMSS1V6CKKB","clice","01M6BZEJN97S3AQVRQ4V7GGE5C",1383,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fclice-io\u002Fclice\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fcodex\u002FSKILL.md",[],[5,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36],"read","running","when","scoped","attacks","gpt-6","self","quota","implementation","only","review","reviewer","plan","gpt","debugging","pr","codex cli","cli","2026-11-10 10:12:30",false,"---\nname: codex\ndescription: Read before running the codex CLI (GPT-6 astra) — the additional reviewer of the pr skill's self-review, plan attacks, debugging, scoped implementation. Only when codex is logged in and has quota.\n---\n\n# Codex Delegation\n\n`codex` is an installed CLI agent backed by GPT-6 astra. Its value is\nindependence: it did not write what it reviews. It is optional — an additional\nreviewer beside the Opus reviewers of the pr skill, a second opinion on a\nplan, a debugger, an implementer of a well-scoped task. Use it only when it\nis logged in (`codex login status`) and has quota; when either fails, carry\non without it and without comment. The one exception is the triage skill,\nwhich stops instead: its classification never leaves codex's sandbox.\n\n## Invocation\n\n```bash\ncodex exec -m gpt-6-astra -c model_reasoning_effort=xhigh \\\n  --dangerously-bypass-approvals-and-sandbox \\\n  -o \u002Ftmp\u002Fcodex-\u003Ctopic>.md \\\n  \"\u003Cprompt>\" \u003C \u002Fdev\u002Fnull\n```\n\n- Always pass `-o` — it writes the final reply to a file; stdout mixes it into\n  the transcript and truncates easily. `-o` holds only the last assistant\n  message: never point it at a file the prompt asks codex to write its report\n  to (or one it might guess) — the message overwrites the report. When the\n  prompt names a report file, give `-o` a clearly different name\n  (`*-final-message.md`).\n- The full bypass is deliberate: the sandbox breaks builds and tooling. Codex\n  therefore runs with your permissions — scope the prompt accordingly.\n- xhigh runs take minutes to tens of minutes: run in the background and keep\n  working, no sleep polling. A background `codex exec` must get `\u003C \u002Fdev\u002Fnull`:\n  with a non-TTY stdin it reads the pipe for extra input and blocks forever\n  (the prompt argument does not help). The same applies to Node — spawn it\n  with `stdio: [\"ignore\", ...]`. Right after launch, `head` the log and\n  confirm the `session id` header appeared; a live process alone proves\n  nothing.\n- The startup header prints `session id: \u003Cuuid>` — capture it whenever a\n  follow-up round is plausible.\n- If `gpt-6-astra` is rejected (plan\u002Fauth), drop `-m` to use the account\n  default, and say so when reporting results.\n- `-c service_tier=fast` switches the run to the fast service tier; use it\n  when the maintainer asks for speed (long migrations, translation passes).\n- Prompt shape: the task, the exact files\u002Fcommands in scope, and the answer\n  format you want (e.g. \"numbered findings, each with a minimal\n  counterexample\"). Codex reads files itself — point at paths instead of\n  pasting content.\n- Codex does not auto-load `.claude\u002F` docs — it discovers only `AGENTS.md`,\n  which this repo does not have. Any run that should follow project rules\n  (review, test writing, implementation) must be told in the prompt which\n  rule files to read first, e.g. `.claude\u002FCLAUDE.md` and the cpp-style skill.\n\nThe review command for the pr skill's self-review loads the repo rules and\ncarries the same fixed constraints as the Opus reviewers:\n\n```bash\ncodex exec -m gpt-6-astra -c model_reasoning_effort=xhigh \\\n  --dangerously-bypass-approvals-and-sandbox -o \u002Ftmp\u002Fcodex-review-\u003Ctopic>.md \\\n  \"Read .claude\u002FCLAUDE.md, .claude\u002Fskills\u002Fcpp-style\u002FSKILL.md and, when\ntest code changed, .claude\u002Fskills\u002Fwrite-tests\u002FSKILL.md, and apply their\nrules. Review the changes in 'git diff origin\u002Fmain...HEAD' for correctness,\nrule violations and test coverage. Report only defects that lead to a\nconcrete failure scenario — the input or state, and the wrong result it\nproduces — each with file:line. Nothing outside this diff's scope. No\nrefactoring, restructuring or style-preference suggestions. No findings is\na valid answer.\" \u003C \u002Fdev\u002Fnull\n```\n\n`codex exec review --base origin\u002Fmain` cannot take a prompt (`--base` and the\nprompt argument are mutually exclusive), so it never sees the repo rules — not\nfor self-review.\n\n## Multi-round sessions\n\n`codex exec resume \u003Csession-id> \"\u003Cfollow-up>\"` continues with full context\n(`--last` picks the newest session). Use it for successive adversarial rounds,\n\"now fix what you found\", or clarifying questions — never restate context in a\nfresh session. Execution-scoped flags are NOT inherited from the resumed\nsession: repeat `-m gpt-6-astra`, `-c model_reasoning_effort=xhigh`,\n`--dangerously-bypass-approvals-and-sandbox`, and a fresh `-o` path on every\nresume, or the follow-up silently runs on the default model at default\neffort, sandboxed, and without an output file. `codex exec fork \u003Csession-id>`\nbranches one history into independent continuations.\n\n## Wording\n\nPrompts and everything codex reads go through OpenAI's cybersecurity\nfilter, which judges phrasing, not intent, and kills the session mid-run\n(the content stays in that session's history, so `resume` trips it again —\nonly a fresh session recovers). Rules:\n\n- Use QA vocabulary: \"review\", \"latent correctness defects\", \"regression\n  check\". Never \"attack surface\", \"hunt\u002Fdig for bugs\", \"fabricate the\n  conditions for the bug\"; a negative control is \"a control variant expected\n  to fail\". Open with the fact that this is routine quality work on our own\n  open-source project (the repo's git config and LICENSE back it up).\n- Do not feed codex internal notes written in hunting slang — extract the\n  scope into a neutral file and keep the ledger edits on this side. Ask for\n  terse, report-only output: long analytical prose triggers the filter too.\n- Successive rounds `resume` the same session (or list the prior findings\n  as exclusions), or the round re-reports what is already known.\n\n## Discipline\n\n- **Codex output is hypothesis, not verdict.** Every concrete claim (\"this\n  input breaks it\") gets an empirical probe before you act on it; \"looks fine\"\n  carries no weight. Codex has correctly refuted arguments this side was sure\n  of, and confidently asserted things a probe then disproved. The probe\n  decides, never authority.\n- **Adversarial loop** (plans\u002Fdesigns): write the doc → codex attacks it\n  (demand concrete counterexamples, not general commentary) → probe each\n  counterexample → revise the doc, recording adopted and refuted findings →\n  `resume` the session for the next round. Stop when a round yields no new\n  confirmed finding.\n- **When codex edits code** (implementation, debug fixes, new tests): review\n  its diff as you would a PR — you own what gets committed. Verification\n  (build + suites) happens in the main session, and the hard rules apply\n  unchanged to codex-authored code. Any run that may modify files gets its\n  own git worktree — the main checkout is for analysis-only runs, or edits\n  will race with this session's.\n- **Codex builds and runs suites on a box, never on this machine** — its\n  prompt has it read `.claude\u002Fskills\u002Fbuild\u002FSKILL.md` and\n  `.claude\u002Fskills\u002Ftest\u002FSKILL.md` first. Never let it run the cases on a\n  box this session might also run them on: two full runs at once slow\n  both. Either codex runs them and you don't, or codex analyzes and you\n  verify.\n- **Keep codex's raw output out of the main context.** Progress checks on a\n  background run (log tail, new commits, report landed?) go through an Opus\n  subagent that returns a digest; a codex-authored diff gets the pr skill's\n  Opus reviewers. The main session spot-checks and decides.\n\n## Recipes\n\n- **Plan review**: point it at the doc path; ask for attacks ranked by\n  severity, each with a minimal counterexample. Fold confirmed findings back\n  into the doc.\n- **Code review**: the review command above, beside the pr skill's Opus\n  reviewers.\n- **Debug**: give the failing test, the repro command, and the suspect area;\n  ask for a root-cause hypothesis plus the experiment that would confirm it.\n  Let it run the repro itself.\n- **Test writing**: point it at the write-tests skill, `tests\u002FREADME.md`\n  and 2-3 neighboring cases as the template; ask it to add cases probing a\n  specific behavior and report which outputs look wrong versus expected.\n  Suspicious snapshot diffs are findings — never `UPDATE_SNAPSHOTS` over them.\n- **Implementation**: a well-scoped task with acceptance criteria and pointers\n  to the 2-3 existing features whose structure it should copy. Then review and\n  verify as above.\n\n## Recovery\n\nTo stop a hung run, kill the PID you recorded at launch (`$!` for a shell\nbackground job, the task's PID otherwise) and its subtree — other codex\nruns may be live in parallel. When the PID is lost,\n`pgrep -af 'codex exe[c]'` lists the candidates to pick from; never\n`pkill -f 'codex exec'`, which matches your own shell's command line and\nkills it.\n\nIf a run dies before writing `-o`, the transcript is at\n`~\u002F.codex\u002Fsessions\u002FYYYY\u002FMM\u002FDD\u002F*.jsonl`; the final reply is the last record\nwith payload `type == \"message\"` and `role == \"assistant\"`. These transcripts\npersist indefinitely and record full prompts, file contents, and command\noutput — treat `~\u002F.codex\u002Fsessions\u002F` as sensitive local data.\n","# Codex Delegation\n\n`codex` is an installed CLI agent backed by GPT-6 astra. Its value is\nindependence: it did not write what it reviews. It is optional — an additional\nreviewer beside the Opus reviewers of the pr skill, a second opinion on a\nplan, a debugger, an implementer of a well-scoped task. Use it only when it\nis logged in (`codex login status`) and has quota; when either fails, carry\non without it and without comment. The one exception is the triage skill,\nwhich stops instead: its classification never leaves codex's sandbox.\n\n## Invocation\n\n```bash\ncodex exec -m gpt-6-astra -c model_reasoning_effort=xhigh \\\n  --dangerously-bypass-approvals-and-sandbox \\\n  -o \u002Ftmp\u002Fcodex-\u003Ctopic>.md \\\n  \"\u003Cprompt>\" \u003C \u002Fdev\u002Fnull\n```\n\n- Always pass `-o` — it writes the final reply to a file; stdout mixes it into\n  the transcript and truncates easily. `-o` holds only the last assistant\n  message: never point it at a file the prompt asks codex to write its report\n  to (or one it might guess) — the message overwrites the report. When the\n  prompt names a report file, give `-o` a clearly different name\n  (`*-final-message.md`).\n- The full bypass is deliberate: the sandbox breaks builds and tooling. Codex\n  therefore runs with your permissions — scope the prompt accordingly.\n- xhigh runs take minutes to tens of minutes: run in the background and keep\n  working, no sleep polling. A background `codex exec` must get `\u003C \u002Fdev\u002Fnull`:\n  with a non-TTY stdin it reads the pipe for extra input and blocks forever\n  (the prompt argument does not help). The same applies to Node — spawn it\n  with `stdio: [\"ignore\", ...]`. Right after launch, `head` the log and\n  confirm the `session id` header appeared; a live process alone proves\n  nothing.\n- The startup header prints `session id: \u003Cuuid>` — capture it whenever a\n  follow-up round is plausible.\n- If `gpt-6-astra` is rejected (plan\u002Fauth), drop `-m` to use the account\n  default, and say so when reporting results.\n- `-c service_tier=fast` switches the run to the fast service tier; use it\n  when the maintainer asks for speed (long migrations, translation passes).\n- Prompt shape: the task, the exact files\u002Fcommands in scope, and the answer\n  format you want (e.g. \"numbered findings, each with a minimal\n  counterexample\"). Codex reads files itself — point at paths instead of\n  pasting content.\n- Codex does not auto-load `.claude\u002F` docs — it discovers only `AGENTS.md`,\n  which this repo does not have. Any run that should follow project rules\n  (review, test writing, implementation) must be told in the prompt which\n  rule files to read first, e.g. `.claude\u002FCLAUDE.md` and the cpp-style skill.\n\nThe review command for the pr skill's self-review loads the repo rules and\ncarries the same fixed constraints as the Opus reviewers:\n\n```bash\ncodex exec -m gpt-6-astra -c model_reasoning_effort=xhigh \\\n  --dangerously-bypass-approvals-and-sandbox -o \u002Ftmp\u002Fcodex-review-\u003Ctopic>.md \\\n  \"Read .claude\u002FCLAUDE.md, .claude\u002Fskills\u002Fcpp-style\u002FSKILL.md and, when\ntest code changed, .claude\u002Fskills\u002Fwrite-tests\u002FSKILL.md, and apply their\nrules. Review the changes in 'git diff origin\u002Fmain...HEAD' for correctness,\nrule violations and test coverage. Report only defects that lead to a\nconcrete failure scenario — the input or state, and the wrong result it\nproduces — each with file:line. Nothing outside this diff's scope. No\nrefactoring, restructuring or style-preference suggestions. No findings is\na valid answer.\" \u003C \u002Fdev\u002Fnull\n```\n\n`codex exec review --base origin\u002Fmain` cannot take a prompt (`--base` and the\nprompt argument are mutually exclusive), so it never sees the repo rules — not\nfor self-review.\n\n## Multi-round sessions\n\n`codex exec resume \u003Csession-id> \"\u003Cfollow-up>\"` continues with full context\n(`--last` picks the newest session). Use it for successive adversarial rounds,\n\"now fix what you found\", or clarifying questions — never restate context in a\nfresh session. Execution-scoped flags are NOT inherited from the resumed\nsession: repeat `-m gpt-6-astra`, `-c model_reasoning_effort=xhigh`,\n`--dangerously-bypass-approvals-and-sandbox`, and a fresh `-o` path on every\nresume, or the follow-up silently runs on the default model at default\neffort, sandboxed, and without an output file. `codex exec fork \u003Csession-id>`\nbranches one history into independent continuations.\n\n## Wording\n\nPrompts and everything codex reads go through OpenAI's cybersecurity\nfilter, which judges phrasing, not intent, and kills the session mid-run\n(the content stays in that session's history, so `resume` trips it again —\nonly a fresh session recovers). Rules:\n\n- Use QA vocabulary: \"review\", \"latent correctness defects\", \"regression\n  check\". Never \"attack surface\", \"hunt\u002Fdig for bugs\", \"fabricate the\n  conditions for the bug\"; a negative control is \"a control variant expected\n  to fail\". Open with the fact that this is routine quality work on our own\n  open-source project (the repo's git config and LICENSE back it up).\n- Do not feed codex internal notes written in hunting slang — extract the\n  scope into a neutral file and keep the ledger edits on this side. Ask for\n  terse, report-only output: long analytical prose triggers the filter too.\n- Successive rounds `resume` the same session (or list the prior findings\n  as exclusions), or the round re-reports what is already known.\n\n## Discipline\n\n- **Codex output is hypothesis, not verdict.** Every concrete claim (\"this\n  input breaks it\") gets an empirical probe before you act on it; \"looks fine\"\n  carries no weight. Codex has correctly refuted arguments this side was sure\n  of, and confidently asserted things a probe then disproved. The probe\n  decides, never authority.\n- **Adversarial loop** (plans\u002Fdesigns): write the doc → codex attacks it\n  (demand concrete counterexamples, not general commentary) → probe each\n  counterexample → revise the doc, recording adopted and refuted findings →\n  `resume` the session for the next round. Stop when a round yields no new\n  confirmed finding.\n- **When codex edits code** (implementation, debug fixes, new tests): review\n  its diff as you would a PR — you own what gets committed. Verification\n  (build + suites) happens in the main session, and the hard rules apply\n  unchanged to codex-authored code. Any run that may modify files gets its\n  own git worktree — the main checkout is for analysis-only runs, or edits\n  will race with this session's.\n- **Codex builds and runs suites on a box, never on this machine** — its\n  prompt has it read `.claude\u002Fskills\u002Fbuild\u002FSKILL.md` and\n  `.claude\u002Fskills\u002Ftest\u002FSKILL.md` first. Never let it run the cases on a\n  box this session might also run them on: two full runs at once slow\n  both. Either codex runs them and you don't, or codex analyzes and you\n  verify.\n- **Keep codex's raw output out of the main context.** Progress checks on a\n  background run (log tail, new commits, report landed?) go through an Opus\n  subagent that returns a digest; a codex-authored diff gets the pr skill's\n  Opus reviewers. The main session spot-checks and decides.\n\n## Recipes\n\n- **Plan review**: point it at the doc path; ask for attacks ranked by\n  severity, each with a minimal counterexample. Fold confirmed findings back\n  into the doc.\n- **Code review**: the review command above, beside the pr skill's Opus\n  reviewers.\n- **Debug**: give the failing test, the repro command, and the suspect area;\n  ask for a root-cause hypothesis plus the experiment that would confirm it.\n  Let it run the repro itself.\n- **Test writing**: point it at the write-tests skill, `tests\u002FREADME.md`\n  and 2-3 neighboring cases as the template; ask it to add cases probing a\n  specific behavior and report which outputs look wrong versus expected.\n  Suspicious snapshot diffs are findings — never `UPDATE_SNAPSHOTS` over them.\n- **Implementation**: a well-scoped task with acceptance criteria and pointers\n  to the 2-3 existing features whose structure it should copy. Then review and\n  verify as above.\n\n## Recovery\n\nTo stop a hung run, kill the PID you recorded at launch (`$!` for a shell\nbackground job, the task's PID otherwise) and its subtree — other codex\nruns may be live in parallel. When the PID is lost,\n`pgrep -af 'codex exe[c]'` lists the candidates to pick from; never\n`pkill -f 'codex exec'`, which matches your own shell's command line and\nkills it.\n\nIf a run dies before writing `-o`, the transcript is at\n`~\u002F.codex\u002Fsessions\u002FYYYY\u002FMM\u002FDD\u002F*.jsonl`; the final reply is the last record\nwith payload `type == \"message\"` and `role == \"assistant\"`. These transcripts\npersist indefinitely and record full prompts, file contents, and command\noutput — treat `~\u002F.codex\u002Fsessions\u002F` as sensitive local data.\n",{"name":5,"description":7},"main","ff37a8a321b684fde2cf6b228bb7ed735f85f30d",".claude\u002Fskills\u002Fcodex",89,"https:\u002F\u002Fgithub.com\u002Fclice-io\u002Fclice","https:\u002F\u002Fdocs.clice.io\u002Fclice\u002F","https:\u002F\u002Fgithub.com\u002Fclice-io","https:\u002F\u002Fclice.io\u002F","2026-11-15 03:05:59",{"source":8,"skillName":5,"commands":52,"prompt":62,"zipAvailable":63},[53,56,59],{"tool":54,"command":55},"npx","npx skills add clice-io\u002Fclice --skill codex",{"tool":57,"command":58},"bunx","bunx skills add clice-io\u002Fclice --skill codex",{"tool":60,"command":61},"pnpm","pnpm dlx skills add clice-io\u002Fclice --skill codex","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fclice-io\u002Fclice\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fcodex\u002FSKILL.md",true,[],[66,94,114,133,152,174],{"skillNo":67,"name":68,"nameZh":6,"description":69,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":70,"categories":71,"keywords":82,"publishedAt":37},"01M6BZFHCK196E4Z66DGN9QDYJ","write-tests","Read before writing or modifying anything under tests\u002F or unittests\u002F, or the tools' own tests — the case workflow, unit tests, hard rules and known pitfalls.","https:\u002F\u002Fgithub.com\u002Fclice-io\u002Fclice\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fwrite-tests\u002FSKILL.md",[72,77],{"categoryNo":73,"categoryName":74,"taxonomyType":75,"taxonomyTypeDesc":76},"01JBSEEDCAT000000000000018","写作与文案","SCENARIO","使用场景",{"categoryNo":78,"categoryName":79,"taxonomyType":80,"taxonomyTypeDesc":81},"01JBSEEDCAT000000000000036","内容创作者","PROFESSION","适用职业",[83,84,85,86,19,87,88,89,90,91,92,93],"write tests","tests","write","pitfalls","case","unit","unit tests","Tools","workflow","Rules","writing",{"skillNo":95,"name":96,"nameZh":6,"description":97,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":98,"categories":99,"keywords":100,"publishedAt":37},"01M6BZFEQ2NCYCKQX0RTMQM55A","upgrade-llvm","Read before upgrading xclang\u002FLLVM (arg = the xclang version, e.g. 23.1.2.1), touching MODULE.bazel, .bazelrc, BUILD.bazel or bazel\u002F, or regenerating modules\u002F — after a library bump, or when clice needs a header or macro it does not export.","https:\u002F\u002Fgithub.com\u002Fclice-io\u002Fclice\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fupgrade-llvm\u002FSKILL.md",[],[101,102,103,104,105,106,19,107,108,21,109,110,111,112,113],"upgrade llvm","llvm","upgrade","export","it","library","build","version","module","bump","modules","bazel","macro",{"skillNo":115,"name":116,"nameZh":6,"description":117,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":118,"categories":119,"keywords":120,"publishedAt":37},"01M6BZFCM31QYASHF2XF6G105C","triage","Use when asked to triage issues — classifies untriaged issues with codex, validates the labels, and returns proposed labels plus an activity digest. Read-only; runs in a forked context.","https:\u002F\u002Fgithub.com\u002Fclice-io\u002Fclice\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Ftriage\u002FSKILL.md",[],[116,121,122,123,124,19,125,126,127,21,128,5,28,129,130,131,132],"fork","context","activity","triage issues","digest","returns","with","use","labels","read-only","issues","activity digest",{"skillNo":134,"name":135,"nameZh":6,"description":136,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":137,"categories":138,"keywords":139,"publishedAt":37},"01M6BZFAW6TTR0FK91HZ0TVNFS","translate-docs","Read before translating, reviewing or editing any docs\u002Fzh page — clice's glossary and page conventions, on top of the shared translation rules of clice-io\u002Fdocs.","https:\u002F\u002Fgithub.com\u002Fclice-io\u002Fclice\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Ftranslate-docs\u002FSKILL.md",[],[140,141,142,143,144,145,19,146,147,148,149,150,92,151],"docs","translate","translate docs","conventions","io","zh","top","translation","shared","editing","page","glossary",{"skillNo":153,"name":154,"nameZh":6,"description":155,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":156,"categories":157,"keywords":161,"publishedAt":37},"01M6BZF86F54V9BZCW3MJW82VY","test","Run the clice test suites, on a box unless a local run is truly necessary; also how to test on Windows. Optional args = suite (unit | cases | all, default all) build type (Debug | RelWithDebInfo, default RelWithDebInfo) and the box id the build ran on. Runs in a forked context; only a digest returns.","https:\u002F\u002Fgithub.com\u002Fclice-io\u002Fclice\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Ftest\u002FSKILL.md",[158],{"categoryNo":159,"categoryName":160,"taxonomyType":75,"taxonomyTypeDesc":76},"01JBSEEDCAT000000000000008","测试与质量",[154,162,88,126,163,164,165,166,167,168,169,122,170,125,107,121,171,172,28,173],"default","run","how","debug","box","windows","local","local run","suite","id","type","how to",{"skillNo":175,"name":176,"nameZh":6,"description":177,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":178,"categories":179,"keywords":180,"publishedAt":37},"01M6BZF5WRJ6PKYJVW4YKPYD20","resolve-comments","Use in each watch round of an open PR — pulls its unresolved review threads, fixes them in the worktree, resolves them, and returns a compact summary.","https:\u002F\u002Fgithub.com\u002Fclice-io\u002Fclice\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fresolve-comments\u002FSKILL.md",[],[181,182,128,183,184,185,186,187,126,29,188,189,34,190],"comments","resolve comments","watch","compact","summary","fixes","threads","open pr","worktree","open"]