agentsclimarketplace

Review

Skill franciscpd/batuta/skills/review

Batuta's on-demand verification. Use when the user invokes /batuta:review or asks to review a diff, an executor's work, or uncommitted changes.From its SKILL.md

Install
npx -y skills add franciscpd/batuta --skill review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.2 KB, as published. Nobody here has run it

Batuta — on-demand review

Re-runs the cycle's verification step over any diff, delegating no code work.

With superpowers installed, conduct the review with the rigor of requesting-code-review and verification-before-completion (plugin root superpowers.md, verification row); the steps and the verdict below stay unchanged.

With the codex plugin installed, also request a Codex second-opinion review of the diff (codex-plugin.md, cross-review row); the verdict below stays the maestro's.

The hardened-verification rules in verification.md apply here in full: report ≠ evidence, test-hygiene scans when the diff touches tests, slop checklist inside the diff review, and the cross-review contract whenever a second reviewer is dispatched.

  1. Determine the target: by default, uncommitted changes (git diff + git diff --staged); the user may point to a range (HEAD~3..), a branch or a commit.
  2. Diff review — review as the maestro: correctness, scope, adherence to the conventions in .batuta/profile.md and the stack template. When a brief with a Scope list is associated, run the scope check first: the target's changed paths (git status --porcelain for uncommitted work, git diff --name-only <range> otherwise) against the list — out-of-list paths are findings. Traceability test: every changed line must trace to what the change set out to do — flag drive-by edits even when correct.
  3. Tests — run the profile's test command (if there is no profile, ask which command to use).
  4. Acceptance criteria — if a brief or plan is associated with the change (the task's run trail in .batuta/runs/ carries the brief verbatim), check item by item; otherwise derive criteria from what the change appears to deliver and make that explicit.
  5. Verdict — ✅ approved or ❌ rejected, with an objective list of issues (file:line) and a suggested next step (fix via /batuta, commit, discard).

This skill neither commits nor changes code — it only evaluates and reports.

Keep looking

Skills are one crate of 610,045. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.