babysit-pr
Use after a pull request is open to watch its CI and review-bot comments until it is ready. Fixes real findings, replies with a…
Applies a refactor plan written by refactor-simple, refactor-deep, or the refactor orchestrator's merged report - auto-fixable items first, then manual items one at a time with the user, verifying with the repo's own checks after each group. Use only after the user has reviewed
$ npx -y skills add greenfield-inc/Pane --skill refactor-apply --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/refactor-applyContext preview
The summary Claude sees to decide when to auto-load this skill.
Applies a refactor plan written by refactor-simple, refactor-deep, or the refactor orchestrator's merged report - auto-fixable items first, then manual items one at a time with the user, verifying with the repo's own checks after each group. Use only after the user has reviewed
name: refactor-apply description: Applies a refactor plan written by refactor-simple, refactor-deep, or the refactor orchestrator's merged report - auto-fixable items first, then manual items one at a time with the user, verifying with the repo's own checks after each group. Use only after the user has reviewed the plan and said to apply it. argument-hint: "<plan-path> [--auto-only]"
Turns a reviewed plan into edits. Never runs unasked; the plan it consumes was shown to the user first, by `refactor` or by the user invoking an analysis directly.
One plan file: an individual `./tmp/simple-refactor-plan-*.md` or `./tmp/deep-refactor-plan-*.md`, or the orchestrator's `./tmp/refactor-merged-*.md`. All three share the same shape - Critical / Warning / Info sections, each item with `file:line`, a fix, and an `Auto-fixable: Yes/No` flag.
Parse every item. Then find what "passing" means in this repository - its lint, typecheck, and test commands from `package.json`, `Makefile`, `CLAUDE.md` or `AGENTS.md`. Run them once before touching anything so a pre-existing failure is not mistaken for one you caused.
Apply every `Auto-fixable: Yes` item, grouped by file. Rules:
A refactor pass does not widen a PR's scope.
auto-fixable - move it to manual and say why.
introduced; revert an item you cannot make pass and report it.
`--auto-only` stops here with a summary.
For each `Auto-fixable: No` item, in plan priority order - Critical first: state the finding, the proposed change, and any judgment call it needs; wait for the user's answer; apply; run the checks. The user may skip any item; a skipped Critical is recorded, not argued.
Applied: X auto-fixable, Y manual (Z skipped by user) Checks: [each command] PASS/FAIL Reverted: [items that could not pass, with reason] Left for follow-up: [skipped or out-of-scope items]
Do not commit. The user (or `prepare-pr`) owns the commit, and the diff should be reviewable as its own step.
user's explicit yes and a test.
user asks for it by name.
Use after a pull request is open to watch its CI and review-bot comments until it is ready. Fixes real findings, replies with a…
Use when a problem, idea, or change of intent is worth capturing as a long-form page, or the person asks to write it up. Never…
Use when the person reports broken behaviour or starts a session to file a bug. Reproduces and reports. Never fixes.
Shared vocabulary for designing deep modules. Use when the user wants to design or improve a module's interface, find deepening…
Explore the codebase to locate files, trace behavior, and document existing patterns with precise file references. Use when the…
Give a fresh agent something to walk through with zero context - a PR body, a landing page, a document, a process, instructions,…