[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7C9RN5GS2046XF50AXR05ZB":3,"skill-same:01M7C9RN5GS2046XF50AXR05ZB":236},{"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":23,"commitSha":24,"skillDirectory":25,"forksCount":13,"repositoryHtmlUrl":26,"repositoryHomepageUrl":6,"ownerHtmlUrl":27,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":28,"install":29,"resources":42},"01M7C9RN5GS2046XF50AXR05ZB","loop-doctor",null,"Reviews — and, when asked, sets up — a project's autonomous, scheduled loop — the routine that wakes on a cadence (cron, launchd, a \u002Floop interval, a cloud routine, a CI schedule, a driver script in a sleep loop, a tick counter), does work unattended, and re-arms. Draws how one tick works from the project's real files, scores the loop on ten dimensions (correctness, safety, reliability, cost, maintainability, understandability, observability, purpose, improvement, proportionality), finds redundant and contradictory rules across the governing files, finds invalid detail (dead references, tripped stop sentinels, competing drivers, unbounded state files, gates the actor can approve itself), and compares the loop against a complete loop concept — works through the roadmap, re-plans from intent when it runs dry, tries-verifies-adjusts until the work is right, challenges finished work against a cited bar, spends in proportion, and uses Jev (if present) for cheap decisions at fixed points. Prescribes what closes each finding; in setup mode writes the loop's rules and config after approval. Use whenever someone asks how their loop \u002F tick \u002F wake \u002F orchestrator \u002F routine works, wants it reviewed, scored, benchmarked, audited, simplified or made safer, says the loop is stuck, drifting, doing busywork or \"doesn't stop\", asks what a scheduled agent actually does each run, or wants to design a periodic autonomous loop and check it before turning it on.","Busy-Office\u002Fai-skills","Busy-Office","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F295484308?v=4","01M7C9Q87FQYDQSSBXBRNMQ1CW","01M7C9Q880V3XY9NB38T9ZCWFG",0,"MIT","https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: loop-doctor\ndescription: Reviews — and, when asked, sets up — a project's autonomous, scheduled loop — the routine that wakes on a cadence (cron, launchd, a \u002Floop interval, a cloud routine, a CI schedule, a driver script in a sleep loop, a tick counter), does work unattended, and re-arms. Draws how one tick works from the project's real files, scores the loop on ten dimensions (correctness, safety, reliability, cost, maintainability, understandability, observability, purpose, improvement, proportionality), finds redundant and contradictory rules across the governing files, finds invalid detail (dead references, tripped stop sentinels, competing drivers, unbounded state files, gates the actor can approve itself), and compares the loop against a complete loop concept — works through the roadmap, re-plans from intent when it runs dry, tries-verifies-adjusts until the work is right, challenges finished work against a cited bar, spends in proportion, and uses Jev (if present) for cheap decisions at fixed points. Prescribes what closes each finding; in setup mode writes the loop's rules and config after approval. Use whenever someone asks how their loop \u002F tick \u002F wake \u002F orchestrator \u002F routine works, wants it reviewed, scored, benchmarked, audited, simplified or made safer, says the loop is stuck, drifting, doing busywork or \"doesn't stop\", asks what a scheduled agent actually does each run, or wants to design a periodic autonomous loop and check it before turning it on.\n---\n\n# Loop Doctor\n\nA scheduled loop runs when nobody is looking, and its design is never in\none file: scheduler → driver → skill → rules in `CLAUDE.md` → a\ndefinition-of-done → a gate log → a state file. Each was written at a\ndifferent time, and they drift. The loop keeps running on whichever copy\nit happens to read.\n\nLoop-doctor has two modes. **Review** (the default) produces one artifact —\na review that is **diagram first, score second, rows third** — and costs the\nproject nothing. **Setup** (only when asked to set up, design or fix a loop)\nwalks through the few decisions the project must make and writes the loop's\nrules and config as one diff, after approval. Both measure against\n`references\u002Floop-concept.md`, the complete loop: roadmap-driven, re-planning\nfrom intent, try-verify-adjust until right, challenged at milestones against\na cited bar, spending in proportion, with Jev at fixed points when present.\n\n```mermaid\nflowchart LR\n  A[\"inventory.mjs\u003Cbr\u002F>≤ 2 s · read-only\"] --> B[\"map onto\u003Cbr\u002F>the tick\"]\n  B --> C[\"audit\u003Cbr\u002F>Invalid · Redundant · Risk\"]\n  C --> D[\"score\u003Cbr\u002F>10 × 0–5 → health\"]\n  D --> E[\"prescribe\u003Cbr\u002F>what closes a finding\"]\n  E --> F[\"review\u003Cbr\u002F>diagram · score · rows · do-next\"]\n  F -. only when asked .-> G[\"apply\"]\n```\n\n## The guard-rail: no overhead on the project\n\nThe person calling loop-doctor is asking a question, not adopting a tool.\nSo:\n\n- **Write nothing into the target repo** in review mode. The review goes to the\n  scratchpad (and to a published page if that's how they read things). If\n  they want it in the repo, they name the path and you write exactly that\n  one file.\n- **Run nothing of the project's.** No install, no build, no tests, no\n  `refresh` command. Only `git log`\u002F`git show`\u002F`git status` and the\n  inventory script.\n- **Read from the digest, not the files.** The inventory's `rulesDigest`\n  (the lines that govern behaviour, with file:line, the sections a tick\n  uses first) and `sections` (every heading's line range, `used` where the\n  loaded rules point) are the reading. Open a file only to confirm a\n  finding's citation or to read a `used` section the digest can't settle —\n  and batch those reads, several ranges in one command. **Budget: ~1,500\n  lines of governing text opened in all** — the rules files, the sections a\n  tick uses, the scripts the rules run, and the intent and config sections;\n  the skill's own files don't count. The receipt reports the number.\n  Cost is context × turns: ten batched reads beat fifty single ones. State\n  files by header, first screen and last screen — the inventory gives their line counts, and\n  the count *is* the finding. Never read a queue or log end to end,\n  unless it is under about 40 lines; say so in the receipt.\n- **Stay blind to earlier reviews.** Don't open the files the inventory\n  lists under `priorReviews`; a previous score anchors the new one. The\n  footer says they were skipped.\n- **No subagents by default.** One context, one pass. The inventory\n  already did the mechanical work.\n- **Leave a receipt.** The review's footer says how many files were read,\n  that no state file was read whole, and that the target repo is\n  untouched — so the reader can see the cost.\n\n## Workflow\n\n### 1. Inventory\n\n```bash\nnode \u003Cskill-dir>\u002Fscripts\u002Finventory.mjs \u003Crepo-path> > \u003Cscratchpad>\u002Floop-inventory.json\n```\n\nTriggers (cron, launchd, workflow `schedule:`, `\u002Floop Nm`, cloud-routine\nmentions), drivers, governing docs, loop skills\u002Fcommands, agents, state\nfiles with size and growth, the intent document and who references it,\nqueue sharpness, and the cheap checks: repeated sentences across files,\ndangling references (dead vs moved), numbers that disagree for the same\nphrase, terminal sentinels with entries after them, swallowed errors,\nsingle-OS drivers. Under `concept`: a loop config file, Jev calls (from\nthe driver or a hook, or only in docs), verifier agents, kill-switch, cap,\ninner-loop and challenge mentions, and milestone tags — the evidence for\nstep 5b. `runway` holds what could stall the loop (re-arm evidence, the\nlock and its stale rule, the oldest open gates, the empty-queue rule) and\nthe recent record commits. Read all of it. It is evidence, not conclusions.\n\nNo trigger, no driver, no loop docs → this project has no scheduled\nloop. Say so in three lines and stop.\n\n### 2. Map onto the tick\n\n`references\u002Ftick-model.md`. Fill the eight nodes — trigger, wake, select,\nact, verify, gate, record, re-arm\u002Fstop — with the project's real file\nnames and numbers. A node with two answers is a finding; a node with none\nis usually the worse one. Answer the three questions once each: who owns\n\"done\", where does state live between ticks, do human gates block or log.\nClimb the empty-queue ladder (steady state → re-plan from intent as\nproposals → bounded explore) and note which rungs exist.\n\nState the **runway** in one line: how long the loop can run unattended\nand what stops it first — a real stop (goal met, steady state, a written\ncap, `HALT`, an escalated repeat failure while other work continues) or a\nstall (a person needed for reversible work, a blocking gate with nothing\nelse to take, an empty queue with no refill rule, no re-arm, a lock with\nno stale rule, \"continue\" typed by hand, a rule the loop skips without\nsaying so). The first stall is the blocker, with file:line; add an\nestimate in ticks or days only when cadence, cap and queue length are all\nwritten down, labelled *estimate*. Check the recent record commits\n(`runway.recentTicks`) against what the selection rule would have picked —\na loop that silently skips its own rule stalls in the worst way: it keeps\nrunning. `tick-model.md` has the lists.\n\nState **built vs declared** in one sentence: what runs today, what the\ndocuments describe in the present tense that is only planned, simulated\nor dormant. In every real loop reviewed so far this was the sentence the\nreader needed most.\n\n### 3. Audit\n\n`references\u002Fchecklist.md` — every item says why it matters unattended and\nhow a miss is classified:\n\n- **Invalid** — the loop does not do what its files say, or cannot.\n- **Redundant** — the same rule in more than one place; name the copy the\n  loop actually loads.\n- **Risk** — nothing wrong yet; the design lets it go wrong silently.\n\nOne row per finding: id · what · where (file:line) · exact fix.\n\"Consider consolidating\" is not a row.\n\nIf the loop has a queue, list the **ambiguous items** — acceptance\nunstated, or judgement-worded (*improve, look at, consider, clean up*) — by\nid and file:line, and hand them to `requeue`. Rewording them is its job.\n\n### 4. Score\n\n`references\u002Fscorecard.md`. Ten dimensions, 0–5 by the anchors, one-line\nreason each citing a file or a number; mean → health label (`fit · watch\n· treat · stop`; a 0 in safety caps at *treat*). The score summarises the\nrows; a reader must be able to go from a score line to the rows behind it.\n\nScore last, and score from the reason: write the one-line reason first —\na file:line or a number from the project, then the finding ids —\ncount the Invalid ids in it, and apply the ceiling (one → ≤ 2, two or more\n→ ≤ 1) before writing the number. Both gauntlet rounds so far drifted one\npoint high on exactly the dimensions whose reasons cited two Invalids.\n\n### 5. Prescribe\n\n`references\u002Fpatterns.md`. Recommend only what closes a finding or a gap\nfrom step 2, with its cost. Three habits kept beat twelve written down.\n\n### 5b. Compare against the loop concept\n\n`references\u002Floop-concept.md` is the target. Go through its slots `c1`–`c26`\nand mark each **present** (file:line), **missing** (where it should live)\nor **contradicted** (both answers). One row for every missing or\ncontradicted slot; the present ones show in the scores. List every **human-input point** (start\nit, answer a blocking gate, clarify an item, refill the queue, notice it\nis stuck, decide it is done) with what removes each. Then the concept's\nrules the loop lacks, **worded for this project's files, paste-ready** for\nthe rules file the driver actually loads — every placeholder a real name,\nand no line that contradicts a rule already in that file. Keep the principles: the human\nleaves the critical path, not the decision; good, not just done — in\nproportion.\n\n### 6. Write the review\n\n`references\u002Freport-template.md`, exactly: ≤ 60 words, the tick diagram\nwith real file names, files-by-role, the score block, the finding tables,\nprescriptions, the concept comparison, do-next (≤ 5), footer receipt. Prose outside tables\u002Fdiagrams\u002Fcode ≤ 450 words in total (the\npaste-ready rule is a code block and doesn't count). Then a terminal\nsummary of ≤ 8 lines: one sentence on the loop, the health label and\nmean, findings by class, the first fix.\n\n### 7. Apply — only when asked\n\nInvalid and Redundant fixes you proposed, one commit per class, diff\nshown first. Risk items and pattern adoption stay separate decisions:\nthey change how the loop behaves, and a loop's behaviour should never\nchange as a side effect of being looked at.\n\n## Setup mode — only when asked\n\nWhen someone asks to set up, design or rebuild their loop (not to review\nit), after steps 1–2:\n\n1. **Ask only what the files can't answer**, each with the concept's default\n   as the recommendation: where the roadmap and intent live (if the\n   inventory didn't find them), which kinds of work need which critique\n   lenses, the per-item and per-week caps, the improvement share, the\n   escalation channel, and whether Jev is available (`jev doctor`). At most\n   seven questions; take \"defaults\" as an answer.\n2. **Draft one diff**, written for this project's files: the concept's\n   rules as a section of the rules file the driver loads (never a second\n   copy); a `loop.config.json` with levels, caps, share, lenses, models,\n   stops, halt file and channel; the role → agent mapping; the Jev calls\n   at their fixed points, each with its fallback, if Jev is available.\n3. **Show it and wait.** Write only after an explicit yes, as one commit.\n   Never write the scheduler or driver entry — say which to use and why;\n   where money and machines are involved, the person decides.\n4. Then run review mode on the result, so the new loop gets the same score\n   as any other.\n\n## Judgement calls\n\n**The copy that's loaded is the real one.** A rule in `CLAUDE.md`, a\n`SKILL.md` and a `docs\u002F*.md` is obeyed in whichever the driver puts in\ncontext. Keep that one; turn the others into a one-line pointer.\n\n**Sentinels are state, not history.** A stop rule that greps for\n`STATUS: COMPLETE` works once. If the log continued past it, the stop\nrule is broken *now*. Rank it above almost everything.\n\n**Growth is a cost paid every tick.** A file read at wake charges its full\nlength to every run. Report lines and growth; if there's no archive rule,\npropose one with a threshold.\n\n**Two drivers is zero drivers.** Don't pick the winner; report that\nprecedence is undefined and propose the person choose.\n\n**Absence of a stop is a finding; absence of a human is not.** Log-and-\ncontinue is a legitimate design. It must then have an asynchronous\nescalation channel and a kill switch — check for those instead.\n\n**Intent is read, never written, by the loop.** Re-planning from intent\nwhen the queue runs dry is the right rung — as gated proposals. A loop\nthat promotes its own proposals is writing its own mandate.\n\n**Good, not just done — in proportion.** A loop with no challenge step\nconverges on \"done\"; one that challenges every item never finishes. Look for\nthe middle: one lens on standard items, a real review per milestone, a\nwritten share and caps.\n\n**The doer never moves the bar.** A loop that can rewrite an acceptance\nline to pass will, eventually. Rank it with self-approval.\n\n**Jev advises; code decides.** A Jev call that lives only in a skill is a\nwish; one that replaces the verifier is a hole. Look for it in the driver\nor a hook, with a fallback, and with its outcomes recorded.\n\n**The owner's decisions stand.** A slot the owner deliberately left out\n(Jev at two points only, a rule they refused) is *not applicable — owner\ndecision* with the file:line, not a gap to argue with. Before pasting or\nprescribing anything, look for a recorded \"refused\", \"do not re-raise\" or\n\"decided\" on the same subject; if there is one, don't paste against it —\none row saying the concept differs and the owner decided, nothing more.\n\n**Continuous means never stalled, not never stopped.** A loop that reaches\nsteady state and stops is healthy; one that always has a next item because\na rule is always true is not continuous, it's unbounded. How clear the\nobjective is belongs to `sharpen-intent`; this review checks the loop is\nanchored to it and knows when it's met.\n\n**No driver script is not no loop.** When the Claude session following the\nrules is the driver, the scripts the rules tell it to run\n(`instructedScripts`) are its code — guards, recorders, locks.\n\n**Say which scale.** The score block names the scale (ten dimensions,\nloop-doctor 0.3). If the inventory's `concept.scoreGates` shows the\nproject gates on a loop-doctor score, give the mean of the first eight\ndimensions too, and say the gate's threshold was set on the eight-dimension\nscale — a silent scale change moves someone's gate.\n\n**Explain before you judge.** If the diagram was hard to draw from the\nfiles, say so once — that difficulty is itself a finding.\n\n## Files\n\n- `scripts\u002Finventory.mjs` — discovery + cheap checks; `--self-test`.\n- `references\u002Ftick-model.md` — the tick, the empty-queue ladder, the\n  three questions.\n- `references\u002Fchecklist.md` — dimensions → checks, with the why.\n- `references\u002Fscorecard.md` — anchors, health label, sharpness.\n- `references\u002Fpatterns.md` — prescriptions catalogue.\n- `references\u002Floop-concept.md` — the target: the complete loop, its roles,\n  Jev points, inner loop, challenge, proportion, doors and stops, config,\n  and paste-ready rules.\n- `references\u002Freport-template.md` — the review, exactly.\n- `evals\u002Fgauntlet\u002F` — the bar this skill's output is graded against.\n\n## Related\n\nIt hands off rather than repeats: ambiguous queue items to `requeue`; a\nvague or missing intent to `sharpen-intent`; what the runs actually cost to\n`loop-economist` (which also runs the concept's weekly cost pass); the\nshape inside one tick to `graph-engineer`; every cheap judgement to `jev`.\n\nFour siblings share this loop's subject and answer different questions:\n`loop-economist` (what the runs actually cost and whether the right agent did\nthe work), `loop-atlas` (what the whole thing looks like, and who is on the\ncrew), `requeue` (what the loop should be fed next) and `sharpen-intent` (what\nthe whole thing is for). This skill judges the **design**: whether the tick is\nsound, safe, and says one thing. When the files are fine but the runs are\nexpensive, it is the economist's question, not this one. And the shape of\nthe work *inside* one tick — how a job fans out across subagents, where the\nverifier sits, whether a barrier is earned — is `graph-engineer`'s subject:\nthis skill reviews the loop around a run, that one the graph within it.\n","# Loop Doctor\n\nA scheduled loop runs when nobody is looking, and its design is never in\none file: scheduler → driver → skill → rules in `CLAUDE.md` → a\ndefinition-of-done → a gate log → a state file. Each was written at a\ndifferent time, and they drift. The loop keeps running on whichever copy\nit happens to read.\n\nLoop-doctor has two modes. **Review** (the default) produces one artifact —\na review that is **diagram first, score second, rows third** — and costs the\nproject nothing. **Setup** (only when asked to set up, design or fix a loop)\nwalks through the few decisions the project must make and writes the loop's\nrules and config as one diff, after approval. Both measure against\n`references\u002Floop-concept.md`, the complete loop: roadmap-driven, re-planning\nfrom intent, try-verify-adjust until right, challenged at milestones against\na cited bar, spending in proportion, with Jev at fixed points when present.\n\n```mermaid\nflowchart LR\n  A[\"inventory.mjs\u003Cbr\u002F>≤ 2 s · read-only\"] --> B[\"map onto\u003Cbr\u002F>the tick\"]\n  B --> C[\"audit\u003Cbr\u002F>Invalid · Redundant · Risk\"]\n  C --> D[\"score\u003Cbr\u002F>10 × 0–5 → health\"]\n  D --> E[\"prescribe\u003Cbr\u002F>what closes a finding\"]\n  E --> F[\"review\u003Cbr\u002F>diagram · score · rows · do-next\"]\n  F -. only when asked .-> G[\"apply\"]\n```\n\n## The guard-rail: no overhead on the project\n\nThe person calling loop-doctor is asking a question, not adopting a tool.\nSo:\n\n- **Write nothing into the target repo** in review mode. The review goes to the\n  scratchpad (and to a published page if that's how they read things). If\n  they want it in the repo, they name the path and you write exactly that\n  one file.\n- **Run nothing of the project's.** No install, no build, no tests, no\n  `refresh` command. Only `git log`\u002F`git show`\u002F`git status` and the\n  inventory script.\n- **Read from the digest, not the files.** The inventory's `rulesDigest`\n  (the lines that govern behaviour, with file:line, the sections a tick\n  uses first) and `sections` (every heading's line range, `used` where the\n  loaded rules point) are the reading. Open a file only to confirm a\n  finding's citation or to read a `used` section the digest can't settle —\n  and batch those reads, several ranges in one command. **Budget: ~1,500\n  lines of governing text opened in all** — the rules files, the sections a\n  tick uses, the scripts the rules run, and the intent and config sections;\n  the skill's own files don't count. The receipt reports the number.\n  Cost is context × turns: ten batched reads beat fifty single ones. State\n  files by header, first screen and last screen — the inventory gives their line counts, and\n  the count *is* the finding. Never read a queue or log end to end,\n  unless it is under about 40 lines; say so in the receipt.\n- **Stay blind to earlier reviews.** Don't open the files the inventory\n  lists under `priorReviews`; a previous score anchors the new one. The\n  footer says they were skipped.\n- **No subagents by default.** One context, one pass. The inventory\n  already did the mechanical work.\n- **Leave a receipt.** The review's footer says how many files were read,\n  that no state file was read whole, and that the target repo is\n  untouched — so the reader can see the cost.\n\n## Workflow\n\n### 1. Inventory\n\n```bash\nnode \u003Cskill-dir>\u002Fscripts\u002Finventory.mjs \u003Crepo-path> > \u003Cscratchpad>\u002Floop-inventory.json\n```\n\nTriggers (cron, launchd, workflow `schedule:`, `\u002Floop Nm`, cloud-routine\nmentions), drivers, governing docs, loop skills\u002Fcommands, agents, state\nfiles with size and growth, the intent document and who references it,\nqueue sharpness, and the cheap checks: repeated sentences across files,\ndangling references (dead vs moved), numbers that disagree for the same\nphrase, terminal sentinels with entries after them, swallowed errors,\nsingle-OS drivers. Under `concept`: a loop config file, Jev calls (from\nthe driver or a hook, or only in docs), verifier agents, kill-switch, cap,\ninner-loop and challenge mentions, and milestone tags — the evidence for\nstep 5b. `runway` holds what could stall the loop (re-arm evidence, the\nlock and its stale rule, the oldest open gates, the empty-queue rule) and\nthe recent record commits. Read all of it. It is evidence, not conclusions.\n\nNo trigger, no driver, no loop docs → this project has no scheduled\nloop. Say so in three lines and stop.\n\n### 2. Map onto the tick\n\n`references\u002Ftick-model.md`. Fill the eight nodes — trigger, wake, select,\nact, verify, gate, record, re-arm\u002Fstop — with the project's real file\nnames and numbers. A node with two answers is a finding; a node with none\nis usually the worse one. Answer the three questions once each: who owns\n\"done\", where does state live between ticks, do human gates block or log.\nClimb the empty-queue ladder (steady state → re-plan from intent as\nproposals → bounded explore) and note which rungs exist.\n\nState the **runway** in one line: how long the loop can run unattended\nand what stops it first — a real stop (goal met, steady state, a written\ncap, `HALT`, an escalated repeat failure while other work continues) or a\nstall (a person needed for reversible work, a blocking gate with nothing\nelse to take, an empty queue with no refill rule, no re-arm, a lock with\nno stale rule, \"continue\" typed by hand, a rule the loop skips without\nsaying so). The first stall is the blocker, with file:line; add an\nestimate in ticks or days only when cadence, cap and queue length are all\nwritten down, labelled *estimate*. Check the recent record commits\n(`runway.recentTicks`) against what the selection rule would have picked —\na loop that silently skips its own rule stalls in the worst way: it keeps\nrunning. `tick-model.md` has the lists.\n\nState **built vs declared** in one sentence: what runs today, what the\ndocuments describe in the present tense that is only planned, simulated\nor dormant. In every real loop reviewed so far this was the sentence the\nreader needed most.\n\n### 3. Audit\n\n`references\u002Fchecklist.md` — every item says why it matters unattended and\nhow a miss is classified:\n\n- **Invalid** — the loop does not do what its files say, or cannot.\n- **Redundant** — the same rule in more than one place; name the copy the\n  loop actually loads.\n- **Risk** — nothing wrong yet; the design lets it go wrong silently.\n\nOne row per finding: id · what · where (file:line) · exact fix.\n\"Consider consolidating\" is not a row.\n\nIf the loop has a queue, list the **ambiguous items** — acceptance\nunstated, or judgement-worded (*improve, look at, consider, clean up*) — by\nid and file:line, and hand them to `requeue`. Rewording them is its job.\n\n### 4. Score\n\n`references\u002Fscorecard.md`. Ten dimensions, 0–5 by the anchors, one-line\nreason each citing a file or a number; mean → health label (`fit · watch\n· treat · stop`; a 0 in safety caps at *treat*). The score summarises the\nrows; a reader must be able to go from a score line to the rows behind it.\n\nScore last, and score from the reason: write the one-line reason first —\na file:line or a number from the project, then the finding ids —\ncount the Invalid ids in it, and apply the ceiling (one → ≤ 2, two or more\n→ ≤ 1) before writing the number. Both gauntlet rounds so far drifted one\npoint high on exactly the dimensions whose reasons cited two Invalids.\n\n### 5. Prescribe\n\n`references\u002Fpatterns.md`. Recommend only what closes a finding or a gap\nfrom step 2, with its cost. Three habits kept beat twelve written down.\n\n### 5b. Compare against the loop concept\n\n`references\u002Floop-concept.md` is the target. Go through its slots `c1`–`c26`\nand mark each **present** (file:line), **missing** (where it should live)\nor **contradicted** (both answers). One row for every missing or\ncontradicted slot; the present ones show in the scores. List every **human-input point** (start\nit, answer a blocking gate, clarify an item, refill the queue, notice it\nis stuck, decide it is done) with what removes each. Then the concept's\nrules the loop lacks, **worded for this project's files, paste-ready** for\nthe rules file the driver actually loads — every placeholder a real name,\nand no line that contradicts a rule already in that file. Keep the principles: the human\nleaves the critical path, not the decision; good, not just done — in\nproportion.\n\n### 6. Write the review\n\n`references\u002Freport-template.md`, exactly: ≤ 60 words, the tick diagram\nwith real file names, files-by-role, the score block, the finding tables,\nprescriptions, the concept comparison, do-next (≤ 5), footer receipt. Prose outside tables\u002Fdiagrams\u002Fcode ≤ 450 words in total (the\npaste-ready rule is a code block and doesn't count). Then a terminal\nsummary of ≤ 8 lines: one sentence on the loop, the health label and\nmean, findings by class, the first fix.\n\n### 7. Apply — only when asked\n\nInvalid and Redundant fixes you proposed, one commit per class, diff\nshown first. Risk items and pattern adoption stay separate decisions:\nthey change how the loop behaves, and a loop's behaviour should never\nchange as a side effect of being looked at.\n\n## Setup mode — only when asked\n\nWhen someone asks to set up, design or rebuild their loop (not to review\nit), after steps 1–2:\n\n1. **Ask only what the files can't answer**, each with the concept's default\n   as the recommendation: where the roadmap and intent live (if the\n   inventory didn't find them), which kinds of work need which critique\n   lenses, the per-item and per-week caps, the improvement share, the\n   escalation channel, and whether Jev is available (`jev doctor`). At most\n   seven questions; take \"defaults\" as an answer.\n2. **Draft one diff**, written for this project's files: the concept's\n   rules as a section of the rules file the driver loads (never a second\n   copy); a `loop.config.json` with levels, caps, share, lenses, models,\n   stops, halt file and channel; the role → agent mapping; the Jev calls\n   at their fixed points, each with its fallback, if Jev is available.\n3. **Show it and wait.** Write only after an explicit yes, as one commit.\n   Never write the scheduler or driver entry — say which to use and why;\n   where money and machines are involved, the person decides.\n4. Then run review mode on the result, so the new loop gets the same score\n   as any other.\n\n## Judgement calls\n\n**The copy that's loaded is the real one.** A rule in `CLAUDE.md`, a\n`SKILL.md` and a `docs\u002F*.md` is obeyed in whichever the driver puts in\ncontext. Keep that one; turn the others into a one-line pointer.\n\n**Sentinels are state, not history.** A stop rule that greps for\n`STATUS: COMPLETE` works once. If the log continued past it, the stop\nrule is broken *now*. Rank it above almost everything.\n\n**Growth is a cost paid every tick.** A file read at wake charges its full\nlength to every run. Report lines and growth; if there's no archive rule,\npropose one with a threshold.\n\n**Two drivers is zero drivers.** Don't pick the winner; report that\nprecedence is undefined and propose the person choose.\n\n**Absence of a stop is a finding; absence of a human is not.** Log-and-\ncontinue is a legitimate design. It must then have an asynchronous\nescalation channel and a kill switch — check for those instead.\n\n**Intent is read, never written, by the loop.** Re-planning from intent\nwhen the queue runs dry is the right rung — as gated proposals. A loop\nthat promotes its own proposals is writing its own mandate.\n\n**Good, not just done — in proportion.** A loop with no challenge step\nconverges on \"done\"; one that challenges every item never finishes. Look for\nthe middle: one lens on standard items, a real review per milestone, a\nwritten share and caps.\n\n**The doer never moves the bar.** A loop that can rewrite an acceptance\nline to pass will, eventually. Rank it with self-approval.\n\n**Jev advises; code decides.** A Jev call that lives only in a skill is a\nwish; one that replaces the verifier is a hole. Look for it in the driver\nor a hook, with a fallback, and with its outcomes recorded.\n\n**The owner's decisions stand.** A slot the owner deliberately left out\n(Jev at two points only, a rule they refused) is *not applicable — owner\ndecision* with the file:line, not a gap to argue with. Before pasting or\nprescribing anything, look for a recorded \"refused\", \"do not re-raise\" or\n\"decided\" on the same subject; if there is one, don't paste against it —\none row saying the concept differs and the owner decided, nothing more.\n\n**Continuous means never stalled, not never stopped.** A loop that reaches\nsteady state and stops is healthy; one that always has a next item because\na rule is always true is not continuous, it's unbounded. How clear the\nobjective is belongs to `sharpen-intent`; this review checks the loop is\nanchored to it and knows when it's met.\n\n**No driver script is not no loop.** When the Claude session following the\nrules is the driver, the scripts the rules tell it to run\n(`instructedScripts`) are its code — guards, recorders, locks.\n\n**Say which scale.** The score block names the scale (ten dimensions,\nloop-doctor 0.3). If the inventory's `concept.scoreGates` shows the\nproject gates on a loop-doctor score, give the mean of the first eight\ndimensions too, and say the gate's threshold was set on the eight-dimension\nscale — a silent scale change moves someone's gate.\n\n**Explain before you judge.** If the diagram was hard to draw from the\nfiles, say so once — that difficulty is itself a finding.\n\n## Files\n\n- `scripts\u002Finventory.mjs` — discovery + cheap checks; `--self-test`.\n- `references\u002Ftick-model.md` — the tick, the empty-queue ladder, the\n  three questions.\n- `references\u002Fchecklist.md` — dimensions → checks, with the why.\n- `references\u002Fscorecard.md` — anchors, health label, sharpness.\n- `references\u002Fpatterns.md` — prescriptions catalogue.\n- `references\u002Floop-concept.md` — the target: the complete loop, its roles,\n  Jev points, inner loop, challenge, proportion, doors and stops, config,\n  and paste-ready rules.\n- `references\u002Freport-template.md` — the review, exactly.\n- `evals\u002Fgauntlet\u002F` — the bar this skill's output is graded against.\n\n## Related\n\nIt hands off rather than repeats: ambiguous queue items to `requeue`; a\nvague or missing intent to `sharpen-intent`; what the runs actually cost to\n`loop-economist` (which also runs the concept's weekly cost pass); the\nshape inside one tick to `graph-engineer`; every cheap judgement to `jev`.\n\nFour siblings share this loop's subject and answer different questions:\n`loop-economist` (what the runs actually cost and whether the right agent did\nthe work), `loop-atlas` (what the whole thing looks like, and who is on the\ncrew), `requeue` (what the loop should be fed next) and `sharpen-intent` (what\nthe whole thing is for). This skill judges the **design**: whether the tick is\nsound, safe, and says one thing. When the files are fine but the runs are\nexpensive, it is the economist's question, not this one. And the shape of\nthe work *inside* one tick — how a job fans out across subagents, where the\nverifier sits, whether a barrier is earned — is `graph-engineer`'s subject:\nthis skill reviews the loop around a run, that one the graph within it.\n",{"name":5,"description":7},"main","ee822ad7e3f1e7dd3305c8e03565dc584906fb76","skills\u002Floop-doctor","https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills","https:\u002F\u002Fgithub.com\u002FBusy-Office","2026-09-27 17:58:58",{"source":8,"skillName":5,"commands":30,"prompt":40,"zipAvailable":41},[31,34,37],{"tool":32,"command":33},"npx","npx skills add Busy-Office\u002Fai-skills --skill loop-doctor",{"tool":35,"command":36},"bunx","bunx skills add Busy-Office\u002Fai-skills --skill loop-doctor",{"tool":38,"command":39},"pnpm","pnpm dlx skills add Busy-Office\u002Fai-skills --skill loop-doctor","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002FSKILL.md",true,[43,49,53,57,61,65,69,73,77,81,85,89,93,97,101,105,109,113,119,124,128,132,136,140,144,148,152,156,160,164,168,172,176,180,184,188,192,196,200,204,208,212,216,220,224,228,232],{"path":44,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":47,"sourceHtmlUrl":48},"evals\u002Fevals.json","REFERENCE","参考文档",4964,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Fevals\u002Fevals.json",{"path":50,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":51,"sourceHtmlUrl":52},"evals\u002Fgauntlet\u002FBAR.md",5955,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Fevals\u002Fgauntlet\u002FBAR.md",{"path":54,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":55,"sourceHtmlUrl":56},"evals\u002Fgauntlet\u002FCRITIC.md",569,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Fevals\u002Fgauntlet\u002FCRITIC.md",{"path":58,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":59,"sourceHtmlUrl":60},"evals\u002Fgauntlet\u002FROUNDS.md",4335,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Fevals\u002Fgauntlet\u002FROUNDS.md",{"path":62,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":63,"sourceHtmlUrl":64},"evals\u002Fgauntlet\u002Frounds\u002Fr2-sample-app-LOOP-REVIEW.md",13567,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Fevals\u002Fgauntlet\u002Frounds\u002Fr2-sample-app-LOOP-REVIEW.md",{"path":66,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":67,"sourceHtmlUrl":68},"evals\u002Fgauntlet\u002Frounds\u002Fr3-sample-app-LOOP-REVIEW.md",14540,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Fevals\u002Fgauntlet\u002Frounds\u002Fr3-sample-app-LOOP-REVIEW.md",{"path":70,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":71,"sourceHtmlUrl":72},"evals\u002Fgauntlet\u002Frounds\u002Fr4-sample-app-LOOP-REVIEW.md",14902,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Fevals\u002Fgauntlet\u002Frounds\u002Fr4-sample-app-LOOP-REVIEW.md",{"path":74,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":75,"sourceHtmlUrl":76},"evals\u002Fgauntlet\u002Frounds\u002Fr5-concept-loop-LOOP-REVIEW.md",10350,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Fevals\u002Fgauntlet\u002Frounds\u002Fr5-concept-loop-LOOP-REVIEW.md",{"path":78,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":79,"sourceHtmlUrl":80},"evals\u002Fgauntlet\u002Frounds\u002Fr6-concept-loop-LOOP-REVIEW.md",16393,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Fevals\u002Fgauntlet\u002Frounds\u002Fr6-concept-loop-LOOP-REVIEW.md",{"path":82,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":83,"sourceHtmlUrl":84},"fixtures\u002Fconcept-loop\u002F.claude\u002Fagents\u002Fverifier.md",106,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fconcept-loop\u002F.claude\u002Fagents\u002Fverifier.md",{"path":86,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":87,"sourceHtmlUrl":88},"fixtures\u002Fconcept-loop\u002FCLAUDE.md",1322,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fconcept-loop\u002FCLAUDE.md",{"path":90,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":91,"sourceHtmlUrl":92},"fixtures\u002Fconcept-loop\u002Fdocs\u002FGATES.md",73,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fconcept-loop\u002Fdocs\u002FGATES.md",{"path":94,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":95,"sourceHtmlUrl":96},"fixtures\u002Fconcept-loop\u002Fdocs\u002FLOOP-LOG.md",61,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fconcept-loop\u002Fdocs\u002FLOOP-LOG.md",{"path":98,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":99,"sourceHtmlUrl":100},"fixtures\u002Fconcept-loop\u002Fdocs\u002FRESUME.md",53,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fconcept-loop\u002Fdocs\u002FRESUME.md",{"path":102,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":103,"sourceHtmlUrl":104},"fixtures\u002Fconcept-loop\u002Fexpect.json",907,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fconcept-loop\u002Fexpect.json",{"path":106,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":107,"sourceHtmlUrl":108},"fixtures\u002Fconcept-loop\u002Fintent.md",193,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fconcept-loop\u002Fintent.md",{"path":110,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":111,"sourceHtmlUrl":112},"fixtures\u002Fconcept-loop\u002FROADMAP.md",507,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fconcept-loop\u002FROADMAP.md",{"path":114,"resourceType":115,"resourceTypeDesc":116,"sizeBytes":117,"sourceHtmlUrl":118},"fixtures\u002Fconcept-loop\u002Fscripts\u002Floop.sh","SCRIPT","脚本",182,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fconcept-loop\u002Fscripts\u002Floop.sh",{"path":120,"resourceType":121,"resourceTypeDesc":122,"sizeBytes":13,"sourceHtmlUrl":123},"fixtures\u002Finstructed-loop\u002F.claude\u002Fscheduled_tasks.lock","OTHER","其他","https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Finstructed-loop\u002F.claude\u002Fscheduled_tasks.lock",{"path":125,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":126,"sourceHtmlUrl":127},"fixtures\u002Finstructed-loop\u002FCLAUDE.md",251,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Finstructed-loop\u002FCLAUDE.md",{"path":129,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":130,"sourceHtmlUrl":131},"fixtures\u002Finstructed-loop\u002Fexpect.json",804,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Finstructed-loop\u002Fexpect.json",{"path":133,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":134,"sourceHtmlUrl":135},"fixtures\u002Finstructed-loop\u002FLOOPS-archive.md",114,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Finstructed-loop\u002FLOOPS-archive.md",{"path":137,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":138,"sourceHtmlUrl":139},"fixtures\u002Finstructed-loop\u002FLOOPS.md",240,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Finstructed-loop\u002FLOOPS.md",{"path":141,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":142,"sourceHtmlUrl":143},"fixtures\u002Finstructed-loop\u002Fnotes\u002FENV.md",24,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Finstructed-loop\u002Fnotes\u002FENV.md",{"path":145,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":146,"sourceHtmlUrl":147},"fixtures\u002Finstructed-loop\u002Fnotes\u002Floop-doctor-rescore-2026-09-20.md",32,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Finstructed-loop\u002Fnotes\u002Floop-doctor-rescore-2026-09-20.md",{"path":149,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":150,"sourceHtmlUrl":151},"fixtures\u002Finstructed-loop\u002Fnotes\u002FRESUME.md",59,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Finstructed-loop\u002Fnotes\u002FRESUME.md",{"path":153,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":154,"sourceHtmlUrl":155},"fixtures\u002Finstructed-loop\u002FROADMAP.md",460,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Finstructed-loop\u002FROADMAP.md",{"path":157,"resourceType":115,"resourceTypeDesc":116,"sizeBytes":158,"sourceHtmlUrl":159},"fixtures\u002Finstructed-loop\u002Fscripts\u002Floops\u002Fguard.py",58,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Finstructed-loop\u002Fscripts\u002Floops\u002Fguard.py",{"path":161,"resourceType":115,"resourceTypeDesc":116,"sizeBytes":162,"sourceHtmlUrl":163},"fixtures\u002Finstructed-loop\u002Fscripts\u002Floops\u002Finflight.py",70,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Finstructed-loop\u002Fscripts\u002Floops\u002Finflight.py",{"path":165,"resourceType":115,"resourceTypeDesc":116,"sizeBytes":166,"sourceHtmlUrl":167},"fixtures\u002Finstructed-loop\u002Fscripts\u002Floops\u002Frecord.py",18,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Finstructed-loop\u002Fscripts\u002Floops\u002Frecord.py",{"path":169,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":170,"sourceHtmlUrl":171},"fixtures\u002Fscript-loop\u002FCLAUDE.md",398,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fscript-loop\u002FCLAUDE.md",{"path":173,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":174,"sourceHtmlUrl":175},"fixtures\u002Fscript-loop\u002Fdocs\u002FBACKLOG.md",539,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fscript-loop\u002Fdocs\u002FBACKLOG.md",{"path":177,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":178,"sourceHtmlUrl":179},"fixtures\u002Fscript-loop\u002Fdocs\u002FDEFINITION-OF-DONE.md",110,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fscript-loop\u002Fdocs\u002FDEFINITION-OF-DONE.md",{"path":181,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":182,"sourceHtmlUrl":183},"fixtures\u002Fscript-loop\u002Fdocs\u002FHUMAN-GATES-LOG.md",162,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fscript-loop\u002Fdocs\u002FHUMAN-GATES-LOG.md",{"path":185,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":186,"sourceHtmlUrl":187},"fixtures\u002Fscript-loop\u002Fdocs\u002FLOOP-STATUS.md",233,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fscript-loop\u002Fdocs\u002FLOOP-STATUS.md",{"path":189,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":190,"sourceHtmlUrl":191},"fixtures\u002Fscript-loop\u002Fdocs\u002FLOOPS.md",263,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fscript-loop\u002Fdocs\u002FLOOPS.md",{"path":193,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":194,"sourceHtmlUrl":195},"fixtures\u002Fscript-loop\u002Fdocs\u002FMANUAL-ACTIONS.md",214,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fscript-loop\u002Fdocs\u002FMANUAL-ACTIONS.md",{"path":197,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":198,"sourceHtmlUrl":199},"fixtures\u002Fscript-loop\u002Fexpect.json",815,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fscript-loop\u002Fexpect.json",{"path":201,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":202,"sourceHtmlUrl":203},"fixtures\u002Fscript-loop\u002Fintent.md",37,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fscript-loop\u002Fintent.md",{"path":205,"resourceType":115,"resourceTypeDesc":116,"sizeBytes":206,"sourceHtmlUrl":207},"fixtures\u002Fscript-loop\u002Fscripts\u002Floop.ps1",299,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Ffixtures\u002Fscript-loop\u002Fscripts\u002Floop.ps1",{"path":209,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":210,"sourceHtmlUrl":211},"references\u002Fchecklist.md",18410,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Freferences\u002Fchecklist.md",{"path":213,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":214,"sourceHtmlUrl":215},"references\u002Floop-concept.md",18549,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Freferences\u002Floop-concept.md",{"path":217,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":218,"sourceHtmlUrl":219},"references\u002Fpatterns.md",7869,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Freferences\u002Fpatterns.md",{"path":221,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":222,"sourceHtmlUrl":223},"references\u002Freport-template.md",5983,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Freferences\u002Freport-template.md",{"path":225,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":226,"sourceHtmlUrl":227},"references\u002Fscorecard.md",4685,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Freferences\u002Fscorecard.md",{"path":229,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":230,"sourceHtmlUrl":231},"references\u002Ftick-model.md",6807,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Freferences\u002Ftick-model.md",{"path":233,"resourceType":115,"resourceTypeDesc":116,"sizeBytes":234,"sourceHtmlUrl":235},"scripts\u002Finventory.mjs",37723,"https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Floop-doctor\u002Fscripts\u002Finventory.mjs",[237,244,250,257,263,270],{"skillNo":238,"name":239,"nameZh":6,"description":240,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":241,"categories":242,"keywords":243,"publishedAt":18},"01M7C9TK43J3W13Q83AHM3ZCWV","loop","one tick","https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fwake-weight\u002Ffixtures\u002Fheavy-wake\u002F.claude\u002Fskills\u002Floop\u002FSKILL.md",[],[],{"skillNo":245,"name":246,"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":247,"categories":248,"keywords":249,"publishedAt":18},"01M7C9TG9ND44SG4X5735RAKVE","wake-weight","https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fwake-weight\u002FSKILL.md",[],[],{"skillNo":251,"name":252,"nameZh":6,"description":253,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":254,"categories":255,"keywords":256,"publishedAt":18},"01M7C9T6K5HNKMW49F9XXAPZVH","solo-flow","Prescribes and audits a trunk-based git workflow for a repository with exactly one writer — a single autonomous agent or loop, with no other humans committing. One branch (main), local-only short-lived work branches, CI as the sole independent check (no PRs, nothing to review socially), and releases as tags cut directly off main rather than a develop\u002Frelease branch dance. Use whenever someone is setting up or reviewing git\u002Fbranching\u002Frelease conventions for an agent-driven or loop-driven repo, asks whether to use GitFlow\u002FGitHub Flow\u002FGitLab Flow\u002Ftrunk-based development, wants AGENTS.md or CLAUDE.md git rules written, is deciding whether a develop branch or long-lived feature branches earn their keep, or reports branch drift, a forgotten back-merge, or an autonomous loop that got confused about which branch it was on. Also use to sanity-check an existing branching setup that looks like GitFlow but has no other human collaborators to justify it, and when deciding how the agent should isolate each unit of work (git worktrees, a primary checkout that never leaves main, running items in parallel).","https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fsolo-flow\u002FSKILL.md",[],[],{"skillNo":258,"name":259,"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":260,"categories":261,"keywords":262,"publishedAt":18},"01M7C9T3HA83YMNZNR8VSM5694","sharpen-intent","https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fsharpen-intent\u002FSKILL.md",[],[],{"skillNo":264,"name":265,"nameZh":6,"description":266,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":267,"categories":268,"keywords":269,"publishedAt":18},"01M7C9SW8YC2TWRYH6BN7NRDH9","requeue","Reviews and reorders the queue an engineering loop or team works from — backlog, roadmap, todo, milestones — so the next work picked up is the right work. Pulls every open item with the attributes that decide order (acceptance stated or not, blocked, human-only, duplicated, contradicted between files, stale), ranks them against the project's intent by unblocking power, purpose, readiness, decay and cost, splits them into lanes (the loop can finish it · needs a bounded subloop · needs a gauntlet of approaches · needs a human), rewrites the top vague items into tasks with an object, a test and a bound, and proposes the new order without touching the repo. Use when someone asks what to work on next, wants the backlog or roadmap reprioritised, re-ranked, triaged or cleaned up, says the priorities are wrong or the roadmap is stale, asks which items the autonomous loop should be fed, or arrives from loop-economist because the loop's problem is its input rather than its actor.","https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Frequeue\u002FSKILL.md",[],[],{"skillNo":271,"name":272,"nameZh":6,"description":273,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":274,"categories":275,"keywords":276,"publishedAt":18},"01M7C9SQXNBEVGRFZR40W24AMW","progress-dashboard","Publishes a one-page stakeholder progress dashboard for any git project — where it stands, what's waiting on people, whether it's stalled — from whatever records the project keeps (ADRs, decision notes, design graphs, roadmaps, gates, backlogs, changelogs, PRs). Use whenever someone asks how a project is going, wants a status update or progress report, asks \"where are we\", \"what's blocked\", \"what's waiting on me\u002Fus\", \"what's left\", or wants an existing dashboard refreshed.","https:\u002F\u002Fgithub.com\u002FBusy-Office\u002Fai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fprogress-dashboard\u002FSKILL.md",[],[]]