agentleFS
Luis85Luis85/agentic-workflow/.claude/skills/review-fix/SKILL.md

review-fix

Skill0 starsChanged 6 months agoMITView on GitHub

Take a single finding from an automated review (the reviewer agent, an external review bot, or a manual review.md line item) and set up a fully isolated workspace for fixing it: a dedicated worktree, a topic branch, and a one‑page plan in docs/plans/. This skill owns plan creation only. The agent then refines the plan, implements via TDD, runs verify, opens a PR, and lets the maintainer (or autonomous‑merge rule) merge. For each finding in scope: The agent (not this…

11 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use review.

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.
  • SentrySkill

    review

    Fixture review skill.

    Also in 5 projects that use this

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 2 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 10 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

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

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

    Run the verify gate (verify).

    line 47
  2. Very specific:

    Wait for review approval + clean state, then merge per feedback_autonomous_merge.md.

    line 53
  3. Distinctive:

    Do not close the source review issue or edit its body.

    line 80
  4. Very specific:

    Do not batch multiple findings into one PR.

    line 81
  5. Very specific:

    Do not force‑create the branch (-B).

    line 82
  6. Distinctive:

    Do not start implementation in this skill.

    line 83

“Common” is measured: kept by at least 6 in 10 of the 10 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

review-fix — turn a review finding into an isolated worktree + plan

Purpose

Take a single finding from an automated review (the reviewer agent, an external review bot, or a manual review.md line item) and set up a fully isolated workspace for fixing it: a dedicated worktree, a topic branch, and a one‑page plan in docs/plans/.

This skill owns plan creation only. The agent then refines the plan, implements via TDD, runs verify, opens a PR, and lets the maintainer (or autonomous‑merge rule) merge.

When to invoke

  • The user names a specific finding ID (e.g. R-API-007, or a sha7‑prefixed <sha7>.<idx> from an external review bot).
  • The user runs /review-fix <id> — single‑finding mode.
  • The user runs /review-fix with no argument — sweep mode: process every still‑open finding from the most recent review.

Pre‑flight checks (refuse if any fail)

  1. The finding ID must exist in the source review artifact (specs/<feature>/review.md, an open tracker issue with the review label, etc.). Refuse free‑text descriptions.
  2. The finding must not already be marked shipped ([x] and a PR/issue reference).
  3. The target worktree path .worktrees/fix-review-<slug> must not already exist. If it does, the prior fix is in flight — do not clobber.
  4. The integration branch must be reachable. git fetch origin first.

How to use

For each finding in scope:

  1. Slugify the finding's headline into kebab‑case, max ~40 chars. Prefix with review- for clarity.
  2. Create the worktree:

    git worktree add .worktrees/fix-review-<slug> \
      -b fix/review-<slug> origin/<integration-branch>
  3. Bootstrap the worktree for the project (npm ci, uv sync, bundle install, …). The verify skill assumes deps are installed.
  4. Scaffold the plan at docs/plans/<YYYY-MM-DD>-review-<slug>.md using templates/idea-template.md adapted for a fix:
    • Finding: the original ID + verbatim quote.
    • Diagnosis: what's wrong, with file:line references.
    • Approach: the smallest change that fixes it.
    • Test plan: the failing test that demonstrates the bug, then the fix.
    • Risk + rollback: what could break, how to revert.
  5. Don't write code yet. Plan only. The agent reviews the plan with the user, then proceeds to TDD.

After the fix is implemented

The agent (not this skill) handles:

  • Run the verify gate (verify).
  • Open the PR; include the canonical magic line in the PR body so the source review can auto‑flip. Format must match exactly (single space after the #<number>):
    Refs #<review-issue-number> finding:<sha7>.<idx>
    This is the same string the review-bot/PROMPT.md "Auto‑flip on merge" section parses. Any deviation (Refs <name>, double space, missing #) breaks the auto‑flip.
  • Wait for review approval + clean state, then merge per feedback_autonomous_merge.md.

Post‑merge cleanup

git worktree remove .worktrees/fix-review-<slug>
git branch -d fix/review-<slug>      # from the main checkout
git worktree prune                   # if needed

Reporting

Single‑finding mode:

Set up fix for <id>:
  worktree: .worktrees/fix-review-<slug>
  branch:   fix/review-<slug>
  plan:     docs/plans/<YYYY-MM-DD>-review-<slug>.md

Next: refine the plan, then implement via TDD inside the worktree.

Sweep mode: one of the above blocks per finding, plus a count summary at the end.

Do not

  • Do not close the source review issue or edit its body. The auto‑flip on merge handles that.
  • Do not batch multiple findings into one PR. One finding, one PR — see feedback_pr_hygiene.md.
  • Do not force‑create the branch (-B). If the branch already exists, the prior fix is in flight.
  • Do not start implementation in this skill. Plan only.

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.