agentleFS
lohi-ailohi-ai/babysit/.claude/skills/triage/SKILL.md

triage

Skill0 starsChanged 3 days agoMITView on GitHub

Tier-1 triage for a stalled or BLOCKED autonomous run. Use when a worker returned BLOCKED/NEEDS_CONTEXT or a ticket's checkpoint stopped advancing — classify recoverable vs needs-human, post a structured handoff, optionally resume from the checkpoint.

190 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use writing-beats, setup-matt-pocock-skills and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • LesaniSkill

    writing-beats

    Shape an article as a journey of beats, choose-your-own-adventure style. The user picks a starting beat from the raw material, you write only that beat, then offer options for where to pivot next, beat by beat, until the article reaches a natural end. Use when the user has raw material and wants to assemble it as a narrative rather than an argument.

    Also in 74 projects that use this

  • GoogleSkill

    setup-matt-pocock-skills

    Configure this repo for the engineering skills — set up its issue tracker, triage label vocabulary, and domain doc layout. Run once before first use of the other engineering skills.

    Also in 182 projects that use this

  • SethGammonSkill

    ascii-diagram

    Generate perfectly aligned ASCII diagrams — architecture, flow, sequence, box-and-arrow. Uses a programmatic character-grid approach so alignment is guaranteed by math, not token prediction. Includes post-render verification.

    Also in 4 projects that use this

  • simo97Skill

    migrate-to-shoehorn

    Migrate test files from `as` type assertions to @total-typescript/shoehorn. Use when user mentions shoehorn, wants to replace `as` in tests, or needs partial test data.

    Also in 74 projects that use this

  • LesaniSkill

    writing-shape

    Take a markdown file of raw material and shape it into an article through a conversational session — drafting candidate openings, growing the piece paragraph by paragraph, arguing about format (lists, tables, callouts, quotes) at each step. Use when the user has a pile of notes, fragments, or a rough draft and wants help turning it into something publishable.

    Also in 70 projects that use this

  • liufeng1976Skill

    setup-ts-deep-modules

    Wire dependency-cruiser into a TypeScript repo so each package is a deep module — implementation hidden in subfolders, reachable only through its entry-point files. User-invoked.

    Also in 47 projects that use this

  • +1 more from Lesani

The file

We split this skill into the rules it gives an agent

1 rule. 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 179 edited versions we read keep these.

  • 0

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use ../semantic-decision/SKILL.md kind recovery with the evidence, available inputs and prior attempts to classify using the criteria below.

    line 21

“Common” is measured: kept by at least 6 in 10 of the 179 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

triage

The layer between mechanical retry and human escalation: read the wreckage, classify recoverable vs needs-human, then restart the run or hand the human a precise ask.

Flow

  1. Resolve the ticket (bbs ticket resolve, or the id from the invocation) and collect the evidence: checkpoint.json (status, step, note), the latest handoff, bbs ticket verdict-status --skill <blocking skill>, the history tail, and the recent clean-handoff-audit rows for this ticket (grep '"clean-handoff-audit"' ~/.babysit/analytics/skill-usage.jsonl | grep '"<ticket>"' | tail) — a run of clean:false with the same issues is the dirty-exit signal Hook C emits but can't fix. No ticket state at all → NEEDS_CONTEXT asking which run to triage.
  2. Name the proximate cause in one sentence, from evidence — never from the status label alone.
  3. Use ../semantic-decision/SKILL.md kind recovery with the evidence, available inputs and prior attempts to classify using the criteria below. Shared refs are filesystem paths beside this skill's directory, so read them by path, not as skill://. Classify:
    • Recoverable — transient environment (network, rate limit, crashed session), mechanical state (dirty tree, stale lock, missing install), or a regenerable artifact (checkpoint says done_step, next step never ran), or an in-scope defect the resumed workflow can repair. Dispatch repair to investigate/implement through the workflow; triage does not edit code. A fresh dispatch from the checkpoint addresses these. A clean-handoff-audit clean:false is a symptom requiring ownership checks: uncommitted changes is the normal handoff (workers never commit), but distinguish this run's work from user/other-run edits before resuming; never adopt all dirt. checkpoint predates the last commit → a commit landed under the run (git flow): verify ticket/head identity, then run bbs autopilot checkpoint --refresh if appropriate and inspect the audit again; refreshing metadata does not prove the work passed.
    • Needs-human — missing input only the human has (credentials, an ambiguous requirement, a product decision), or a finding whose repair needs missing authority or a change to the accepted direction. A genuine but locally repairable finding alone does not require human review.
  4. Post the triage handoff to the ticket (bbs ticket add-handoff --skill triage --status <status> --body-file <path>): cause, classification, prior attempts, evidence pointers and planned action. Persist the retry decision before dispatch so a restart cannot spend it twice; append the observed result afterward.
  5. Act:
    • Recoverable → clear the mechanical cause if trivial (never --force, never delete work), then re-dispatch /bbs:autopilot <workflow> <ticket> — it resumes from the checkpoint on disk. One retry per cause: if a prior triage handoff names the same cause, escalate as needs-human instead of looping.
    • Needs-human → emit the structured NEEDS_CONTEXT block naming the exact missing input and the checkpoint it unblocks.

Rules

  • Triage is diagnosis + dispatch, not repair. A fix that needs code changes is an investigate or implement ticket, not a triage action.
  • Bounded blast radius: no force-push, no deleting branches/worktrees, no dropping the checkpoint.
  • Log the classification to the decision trail (it's a Taste call).

Output

STATUS: DONE | NEEDS_CONTEXT | BLOCKED
CLASSIFICATION: recoverable | needs-human
CAUSE: <one sentence>
ACTION: <re-dispatched from step X | escalated with NEEDS_CONTEXT | none>
NEXT: <what the human should expect or provide>

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.