[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6CJ8SCF1A0BK7J1BKG5ZMPJ":3,"skill-same:01M6CJ8SCF1A0BK7J1BKG5ZMPJ":61},{"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":34,"linkOnly":35,"markdownContent":36,"markdownBody":37,"frontMatter":38,"refName":40,"commitSha":41,"skillDirectory":42,"forksCount":43,"repositoryHtmlUrl":44,"repositoryHomepageUrl":45,"ownerHtmlUrl":46,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":47,"install":48,"resources":60},"01M6CJ8SCF1A0BK7J1BKG5ZMPJ","voice-review",null,"Review the voice-turn comparison log for fidelity drift (what the user said vs what was cleaned, and what the agent said vs what the wingman spoke), archive flagged turns locally, and file sanitized GitHub issues. Triggers on \"\u002Fvoice-review\", \"review voice log\", \"review wingman fidelity\", \"review voice turns\".","thefrederiksen\u002Fdevthrottle","thefrederiksen","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F220024918?v=4","01M6CJ5XSV9624762E0FDS49BZ","01M6CJ5XY3WWF9943901Y87DEX",110,"MIT","https:\u002F\u002Fgithub.com\u002Fthefrederiksen\u002Fdevthrottle\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fvoice-review\u002FSKILL.md",[],[18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33],"review","voice","voice review","github","github issues","issues","file","archive","user","log","what the","drift","triggers","comparison","true","for","2026-11-10 10:12:30",false,"---\nname: voice-review\ndescription: Review the voice-turn comparison log for fidelity drift (what the user said vs what was cleaned, and what the agent said vs what the wingman spoke), archive flagged turns locally, and file sanitized GitHub issues. Triggers on \"\u002Fvoice-review\", \"review voice log\", \"review wingman fidelity\", \"review voice turns\".\ndisable-model-invocation: true\n---\n\n# Voice Review\n\nReview the voice-mode turn log, judge how faithfully each turn was handled, archive\nthe problem turns somewhere personal (NOT the repo), and file content-free GitHub\nissues recommending what to fix and why.\n\nThis skill is the large-language-model judgment layer over the per-turn log added in\nissue #141. It runs nightly via a local Windows Scheduled Task and can be run on demand.\n\n## THE PRIVACY RULE (read first, never break)\n\nEverything the user says by voice is personal. Therefore:\n\n- A GitHub issue MUST NEVER contain recording content: no raw transcript, no cleaned\n  transcript, no agent reply text, no spoken-reply text, no audio, and no quotes or\n  paraphrases of any of them. Not even a snippet.\n- A GitHub issue MAY reference a turn ONLY by its turn id and the archived local file\n  path (which lives outside the repo). The path and id are safe; the content is not.\n- The detailed analysis WITH the actual quotes lives ONLY in the local archive under\n  `%LOCALAPPDATA%\\cc-director\\voice-review\\`, never in the repo and never on GitHub.\n\nIf you cannot describe a defect without quoting the recording, describe the defect\nCLASS in your own abstract words (e.g. \"numeric facts dropped from the spoken reply\")\nand point to the archived turn for the human to inspect privately.\n\n## Data locations\n\n- Source log (auto-purges after 5 days):\n  `%LOCALAPPDATA%\\cc-director\\voice-turn-logs\\\u003Ctimestamp>_\u003CturnId>\\`\n  containing `audio.\u003Cext>`, `inbound.json`, `outbound.json`.\n- Personal archive (created by this skill, persists, OUTSIDE the repo):\n  `%LOCALAPPDATA%\\cc-director\\voice-review\\flagged\\\u003CturnId>\\`  - full copy of a flagged\n  turn plus `review.md` (the detailed analysis with quotes).\n  `%LOCALAPPDATA%\\cc-director\\voice-review\\digests\\\u003CYYYY-MM-DD>.md`  - the run digest.\n\nOn Windows expand `%LOCALAPPDATA%` to `C:\\Users\\\u003Cuser>\\AppData\\Local`. In bash it is\n`$LOCALAPPDATA`.\n\n## What to judge (the rubric)\n\nFor each turn read `inbound.json` and `outbound.json` and assess two hops:\n\n1. INBOUND fidelity - `RawTranscript` vs `CleanedTranscript`:\n   - Did the cleanup change the user's intent, drop a clause, answer\u002Fretract something\n     differently, or over-condense rambling into a different meaning?\n   - Fixing an obvious mishearing or removing filler is GOOD. Changing what was asked\n     is BAD.\n\n2. OUTBOUND fidelity - `AgentReply` vs `WingmanSpoken`:\n   - Were concrete facts dropped (names, numbers, yes\u002Fno, the decision\u002Fresult)?\n   - Was the topic reframed, the answer softened, or the question left unanswered?\n   - Speaking facts ear-friendly (e.g. \"SaaS\" -> \"software as a service\", verbalizing\n     an address) is GOOD. Losing the fact is BAD.\n\nAssign each turn a verdict: `ok` (faithful), `minor` (cosmetic, no meaning lost), or\n`flag` (meaning changed or a fact lost). Record a one-line reason in your own words.\n\n## Steps\n\n1. **Find unreviewed turns.** List `voice-turn-logs\u002F*\u002F`. A turn is unreviewed if it has\n   no `reviewed.json`. Review only those (so re-runs are incremental).\n\n2. **Judge each turn** on the rubric above. Be a strict reader: completeness and\n   intent-preservation are the bar, not vibes.\n\n3. **Archive flagged turns.** For every `flag` (and `minor` if you think it is worth\n   keeping), copy the whole turn directory to\n   `%LOCALAPPDATA%\\cc-director\\voice-review\\flagged\\\u003CturnId>\\` and write a `review.md`\n   there containing your full analysis WITH the relevant quotes (this file is personal\n   and stays local). This is the durable evidence, because the source purges at 5 days.\n\n4. **Write the digest.** Append a section to today's digest at\n   `%LOCALAPPDATA%\\cc-director\\voice-review\\digests\\\u003CYYYY-MM-DD>.md`: how many turns\n   reviewed, the count per verdict, and a one-line-per-flag summary (quotes allowed\n   here - it is local and personal). Keep it skimmable.\n\n5. **Mark reviewed.** Write `reviewed.json` into each source turn dir with\n   `{ verdict, reason, reviewedUtc }` so it is not re-reviewed before it purges.\n\n6. **File GitHub issues - but only for things worth fixing, and SANITIZED.**\n   - Decide what is systematic. A single odd turn is usually evidence to watch, not an\n     issue. A repeated pattern (same defect class across multiple turns) is an issue.\n   - Categorize each as `bug` (a fidelity regression, label `bug`) or `enhancement`\n     (a quality\u002Frobustness improvement, label `enhancement`). Always also add the\n     `voice-review` label (create it once if missing:\n     `gh label create voice-review --description \"Filed by the voice-review skill\" --color BFD4F2`).\n   - DEDUP before creating: `gh issue list --label voice-review --state open`. If an\n     open issue already covers this defect class, add a brief comment that it recurred\n     (referencing new turn ids\u002Fpaths only) instead of opening a duplicate.\n   - Each issue MUST state, content-free:\n       - the defect class (abstract description, no recording content),\n       - WHY it matters (the fidelity consequence to the user),\n       - the RECOMMENDED fix and where (e.g. tighten the inbound cleanup prompt in\n         `src\u002FCcDirector.Core\u002FWingman\u002FWingmanService.cs`, or the outbound prompt in\n         `src\u002FCcDirector.Core\u002FVoice\u002FServices\u002FClaudeSummarizer.cs`),\n       - a REFERENCE block: the turn ids and their archived local paths under\n         `voice-review\u002Fflagged\u002F` so the human can inspect the evidence privately.\n   - Use a HEREDOC for the body. Run the sanitization self-check below before creating.\n\n7. **Report back** (to the console \u002F digest, local only): list issues created or\n   commented, and the archive paths. Quotes are fine here since it is local.\n\n## Sanitization self-check (run before every `gh` call)\n\nBefore creating or commenting on any issue, re-read the exact text you are about to send\nand confirm: it contains NO transcript text, NO agent reply text, NO spoken-reply text,\nand NO paraphrase of any of them - only the abstract defect class, the why, the fix, and\nturn ids + archived paths. If any recording content is present, rewrite it abstractly.\n\n## Issue body template (content-free)\n\n```markdown\n## Defect class\n[Abstract description of the fidelity problem. NO recording content.]\n\n## Why it matters\n[The consequence to the user: meaning changed \u002F fact lost \u002F question unanswered.]\n\n## Recommended fix\n[What to change and where, with file path. Why this fix addresses the class.]\n\n## Evidence (local, personal - not included here)\nTurn ids: \u003Cid1>, \u003Cid2>\nArchived at: %LOCALAPPDATA%\\cc-director\\voice-review\\flagged\\\u003Cid1>\\ (and \u003Cid2>)\nReviewed by the voice-review skill on \u003Cdate>.\n```\n\n## Notes\n\n- The agent running this skill IS the language model doing the judgment; do not shell\n  out to another model. Read the JSON, reason, decide.\n- Never delete source turns yourself; the 5-day purge handles retention. Your archive is\n  the long-lived copy.\n- If there are no unreviewed turns, write a one-line \"nothing new\" digest entry and stop.\n  Do not create issues when there is nothing to report.\n- This skill performs external actions (GitHub issues). When run interactively, show the\n  sanitized issue text before creating it if the user is present.\n","# Voice Review\n\nReview the voice-mode turn log, judge how faithfully each turn was handled, archive\nthe problem turns somewhere personal (NOT the repo), and file content-free GitHub\nissues recommending what to fix and why.\n\nThis skill is the large-language-model judgment layer over the per-turn log added in\nissue #141. It runs nightly via a local Windows Scheduled Task and can be run on demand.\n\n## THE PRIVACY RULE (read first, never break)\n\nEverything the user says by voice is personal. Therefore:\n\n- A GitHub issue MUST NEVER contain recording content: no raw transcript, no cleaned\n  transcript, no agent reply text, no spoken-reply text, no audio, and no quotes or\n  paraphrases of any of them. Not even a snippet.\n- A GitHub issue MAY reference a turn ONLY by its turn id and the archived local file\n  path (which lives outside the repo). The path and id are safe; the content is not.\n- The detailed analysis WITH the actual quotes lives ONLY in the local archive under\n  `%LOCALAPPDATA%\\cc-director\\voice-review\\`, never in the repo and never on GitHub.\n\nIf you cannot describe a defect without quoting the recording, describe the defect\nCLASS in your own abstract words (e.g. \"numeric facts dropped from the spoken reply\")\nand point to the archived turn for the human to inspect privately.\n\n## Data locations\n\n- Source log (auto-purges after 5 days):\n  `%LOCALAPPDATA%\\cc-director\\voice-turn-logs\\\u003Ctimestamp>_\u003CturnId>\\`\n  containing `audio.\u003Cext>`, `inbound.json`, `outbound.json`.\n- Personal archive (created by this skill, persists, OUTSIDE the repo):\n  `%LOCALAPPDATA%\\cc-director\\voice-review\\flagged\\\u003CturnId>\\`  - full copy of a flagged\n  turn plus `review.md` (the detailed analysis with quotes).\n  `%LOCALAPPDATA%\\cc-director\\voice-review\\digests\\\u003CYYYY-MM-DD>.md`  - the run digest.\n\nOn Windows expand `%LOCALAPPDATA%` to `C:\\Users\\\u003Cuser>\\AppData\\Local`. In bash it is\n`$LOCALAPPDATA`.\n\n## What to judge (the rubric)\n\nFor each turn read `inbound.json` and `outbound.json` and assess two hops:\n\n1. INBOUND fidelity - `RawTranscript` vs `CleanedTranscript`:\n   - Did the cleanup change the user's intent, drop a clause, answer\u002Fretract something\n     differently, or over-condense rambling into a different meaning?\n   - Fixing an obvious mishearing or removing filler is GOOD. Changing what was asked\n     is BAD.\n\n2. OUTBOUND fidelity - `AgentReply` vs `WingmanSpoken`:\n   - Were concrete facts dropped (names, numbers, yes\u002Fno, the decision\u002Fresult)?\n   - Was the topic reframed, the answer softened, or the question left unanswered?\n   - Speaking facts ear-friendly (e.g. \"SaaS\" -> \"software as a service\", verbalizing\n     an address) is GOOD. Losing the fact is BAD.\n\nAssign each turn a verdict: `ok` (faithful), `minor` (cosmetic, no meaning lost), or\n`flag` (meaning changed or a fact lost). Record a one-line reason in your own words.\n\n## Steps\n\n1. **Find unreviewed turns.** List `voice-turn-logs\u002F*\u002F`. A turn is unreviewed if it has\n   no `reviewed.json`. Review only those (so re-runs are incremental).\n\n2. **Judge each turn** on the rubric above. Be a strict reader: completeness and\n   intent-preservation are the bar, not vibes.\n\n3. **Archive flagged turns.** For every `flag` (and `minor` if you think it is worth\n   keeping), copy the whole turn directory to\n   `%LOCALAPPDATA%\\cc-director\\voice-review\\flagged\\\u003CturnId>\\` and write a `review.md`\n   there containing your full analysis WITH the relevant quotes (this file is personal\n   and stays local). This is the durable evidence, because the source purges at 5 days.\n\n4. **Write the digest.** Append a section to today's digest at\n   `%LOCALAPPDATA%\\cc-director\\voice-review\\digests\\\u003CYYYY-MM-DD>.md`: how many turns\n   reviewed, the count per verdict, and a one-line-per-flag summary (quotes allowed\n   here - it is local and personal). Keep it skimmable.\n\n5. **Mark reviewed.** Write `reviewed.json` into each source turn dir with\n   `{ verdict, reason, reviewedUtc }` so it is not re-reviewed before it purges.\n\n6. **File GitHub issues - but only for things worth fixing, and SANITIZED.**\n   - Decide what is systematic. A single odd turn is usually evidence to watch, not an\n     issue. A repeated pattern (same defect class across multiple turns) is an issue.\n   - Categorize each as `bug` (a fidelity regression, label `bug`) or `enhancement`\n     (a quality\u002Frobustness improvement, label `enhancement`). Always also add the\n     `voice-review` label (create it once if missing:\n     `gh label create voice-review --description \"Filed by the voice-review skill\" --color BFD4F2`).\n   - DEDUP before creating: `gh issue list --label voice-review --state open`. If an\n     open issue already covers this defect class, add a brief comment that it recurred\n     (referencing new turn ids\u002Fpaths only) instead of opening a duplicate.\n   - Each issue MUST state, content-free:\n       - the defect class (abstract description, no recording content),\n       - WHY it matters (the fidelity consequence to the user),\n       - the RECOMMENDED fix and where (e.g. tighten the inbound cleanup prompt in\n         `src\u002FCcDirector.Core\u002FWingman\u002FWingmanService.cs`, or the outbound prompt in\n         `src\u002FCcDirector.Core\u002FVoice\u002FServices\u002FClaudeSummarizer.cs`),\n       - a REFERENCE block: the turn ids and their archived local paths under\n         `voice-review\u002Fflagged\u002F` so the human can inspect the evidence privately.\n   - Use a HEREDOC for the body. Run the sanitization self-check below before creating.\n\n7. **Report back** (to the console \u002F digest, local only): list issues created or\n   commented, and the archive paths. Quotes are fine here since it is local.\n\n## Sanitization self-check (run before every `gh` call)\n\nBefore creating or commenting on any issue, re-read the exact text you are about to send\nand confirm: it contains NO transcript text, NO agent reply text, NO spoken-reply text,\nand NO paraphrase of any of them - only the abstract defect class, the why, the fix, and\nturn ids + archived paths. If any recording content is present, rewrite it abstractly.\n\n## Issue body template (content-free)\n\n```markdown\n## Defect class\n[Abstract description of the fidelity problem. NO recording content.]\n\n## Why it matters\n[The consequence to the user: meaning changed \u002F fact lost \u002F question unanswered.]\n\n## Recommended fix\n[What to change and where, with file path. Why this fix addresses the class.]\n\n## Evidence (local, personal - not included here)\nTurn ids: \u003Cid1>, \u003Cid2>\nArchived at: %LOCALAPPDATA%\\cc-director\\voice-review\\flagged\\\u003Cid1>\\ (and \u003Cid2>)\nReviewed by the voice-review skill on \u003Cdate>.\n```\n\n## Notes\n\n- The agent running this skill IS the language model doing the judgment; do not shell\n  out to another model. Read the JSON, reason, decide.\n- Never delete source turns yourself; the 5-day purge handles retention. Your archive is\n  the long-lived copy.\n- If there are no unreviewed turns, write a one-line \"nothing new\" digest entry and stop.\n  Do not create issues when there is nothing to report.\n- This skill performs external actions (GitHub issues). When run interactively, show the\n  sanitized issue text before creating it if the user is present.\n",{"name":5,"description":7,"disable-model-invocation":39},true,"main","12032d2a3437cdccc46ff85a50252c60258f2e2e",".claude\u002Fskills\u002Fvoice-review",12,"https:\u002F\u002Fgithub.com\u002Fthefrederiksen\u002Fdevthrottle","https:\u002F\u002Fdevthrottle.com\u002F?ref=gh-about","https:\u002F\u002Fgithub.com\u002Fthefrederiksen","2026-11-11 12:28:38",{"source":8,"skillName":5,"commands":49,"prompt":59,"zipAvailable":39},[50,53,56],{"tool":51,"command":52},"npx","npx skills add thefrederiksen\u002Fdevthrottle --skill voice-review",{"tool":54,"command":55},"bunx","bunx skills add thefrederiksen\u002Fdevthrottle --skill voice-review",{"tool":57,"command":58},"pnpm","pnpm dlx skills add thefrederiksen\u002Fdevthrottle --skill voice-review","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fthefrederiksen\u002Fdevthrottle\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fvoice-review\u002FSKILL.md",[],[62,92,110,137,164,186],{"skillNo":63,"name":64,"nameZh":6,"description":65,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":66,"categories":67,"keywords":68,"publishedAt":34},"01M6CJ9FZRKCB1Y8AY6WAK3JJM","devthrottle-sessions","Talk to the other DevThrottle sessions running across your machines - list them, rename this one, send one of the rare queued messages, read the inbox, reply, open a new session, and close one down. Use when the task involves: message another session, ask another session, list sessions, what sessions are running, spawn a session, rename this session, close this session.","https:\u002F\u002Fgithub.com\u002Fthefrederiksen\u002Fdevthrottle\u002Fblob\u002Fmain\u002Fplugins\u002Fdevthrottle\u002Fskills\u002Fdevthrottle-sessions\u002FSKILL.md",[],[69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91],"sessions","devthrottle sessions","inbox","messages","new session","open","ask","read","list","spawn","session","use","reply","task","rename","message","across","close","down","send","when","running","new",{"skillNo":93,"name":94,"nameZh":6,"description":95,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":96,"categories":97,"keywords":98,"publishedAt":34},"01M6CJ9CYRKE8TSBX67XPTT65R","devthrottle-install","Hand the user the documented steps to install or update DevThrottle, and to connect a gateway afterwards. Use when the task involves: install DevThrottle, set up DevThrottle, update DevThrottle, connect a gateway, cc-devthrottle not found, where do the cc tools come from.","https:\u002F\u002Fgithub.com\u002Fthefrederiksen\u002Fdevthrottle\u002Fblob\u002Fmain\u002Fplugins\u002Fdevthrottle\u002Fskills\u002Fdevthrottle-install\u002FSKILL.md",[],[99,100,101,102,103,104,105,26,80,106,107,108,82,109,89],"install","devthrottle install","connect","update","gateway","SET","Tools","up","cc","do","hand",{"skillNo":111,"name":112,"nameZh":6,"description":113,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":114,"categories":115,"keywords":116,"publishedAt":34},"01M6CJ99Q2QJSXDYAKRBWPGKN2","proof-covers-the-wrong-thing","A proof can be valid and still say nothing about what you changed - wrong surface, wrong reader, wrong caller, or an untested premise. Name what your evidence does NOT cover. Use when the task involves: prove it, by construction, root cause, reproduced it, repro, guard-local, skipped test, suite is green, premise, assumption, self review, it is settled.","https:\u002F\u002Fgithub.com\u002Fthefrederiksen\u002Fdevthrottle\u002Fblob\u002Fmain\u002Fplugins\u002Fagent-discipline\u002Fskills\u002Fproof-covers-the-wrong-thing\u002FSKILL.md",[],[117,118,119,18,120,121,122,123,124,80,125,126,127,82,128,129,130,131,132,133,134,135,136,89],"proof covers the wrong thing","proof","Wrong","valid","test","self review","guard","reader","evidence","repro","say","green","self","local","you","can","cover","suite","prove","root cause",{"skillNo":138,"name":139,"nameZh":6,"description":140,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":141,"categories":142,"keywords":143,"publishedAt":34},"01M6CJ96VERHA4PKRHMF9ZXTBF","destructive-sweeps-lean-to-keep","A destructive operation acts only on what it can positively prove is disposable. Enumerate what to DELETE, never what to skip - and refresh the safety signal on every real use. Use when the task involves: sweep, purge, cleanup, delete old, prune, reclaim space, stale files, retention, aged out, rm -rf, git clean, drop the table.","https:\u002F\u002Fgithub.com\u002Fthefrederiksen\u002Fdevthrottle\u002Fblob\u002Fmain\u002Fplugins\u002Fagent-discipline\u002Fskills\u002Fdestructive-sweeps-lean-to-keep\u002FSKILL.md",[],[144,145,146,147,148,149,150,151,152,153,154,80,155,156,157,158,159,82,160,161,162,163,135,89],"lean","destructive sweeps lean to keep","keep","cleanup","delete","git","Safety","prune","table","sweep","refresh","never","only","stale","files","clean","drop","retention","reclaim space","signal",{"skillNo":165,"name":166,"nameZh":6,"description":167,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":168,"categories":169,"keywords":170,"publishedAt":34},"01M6CJ93GRYMHCZZS657RBHPW6","checks-that-fail-open","A check whose pass condition is an ABSENCE certifies a run that never happened. Restate it as a specific PRESENCE - and an empty result is a broken instrument, never a clean run. Use when the task involves: verify, prove it, sweep, grep returned nothing, zero hits, clean run, it passed, no defects, reviewed, delivered, double check, how do we know.","https:\u002F\u002Fgithub.com\u002Fthefrederiksen\u002Fdevthrottle\u002Fblob\u002Fmain\u002Fplugins\u002Fagent-discipline\u002Fskills\u002Fchecks-that-fail-open\u002FSKILL.md",[],[74,171,172,173,174,175,153,176,177,178,80,155,179,159,108,82,180,181,182,183,184,185,135,89],"checks that fail open","run","ZeRO","verify","how","grep","check","broken","double check","as","condition","restate","presence","defects","result",{"skillNo":187,"name":188,"nameZh":6,"description":189,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":190,"categories":191,"keywords":192,"publishedAt":34},"01M6CJ906AQPKFA2CCKKHAZC13","wingman-brief-reviewer","RETIRED 2026-09-16 - do not run it. It reviewed saved wingman turn BRIEFS, whose writer was retired in issue","https:\u002F\u002Fgithub.com\u002Fthefrederiksen\u002Fdevthrottle\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fwingman-brief-reviewer\u002Fskill.md",[],[193,194,195,172,196,108,197,198,199,200,201],"brief","reviewer","wingman brief reviewer","issue","writer","it","briefs","saved","2026"]