Add /review-workflow skill#2
Open
hefgi wants to merge 4 commits into
Open
Conversation
Mechanical phases (scope, file-listing) run on Haiku; the review ladder converges on Sonnet then re-converges on Opus for the final sign-off, with each round's verify matching its review tier. The no-progress guard resets per tier so a smarter model surfacing new findings counts as progress.
Accept a preset (sonnet-opus, sonnet, opus) or an explicit models= list; when none is given, ask the user with Sonnet -> Opus as the recommended default, plus Sonnet-only and Opus-only. The chosen list becomes the review tiers.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Adds a new project-agnostic skill
/review-workflowto the collection: an iterative review → fix → commit → re-review loop that runs until a code review comes back clean.How it works (two actors)
A Claude Code workflow can orchestrate review subagents but cannot edit the working tree or commit. So the loop is split:
review-workflow/references/review-loop.mjs) — the per-round review engine: scope resolution → parallel dimension reviewers → adversarial per-finding verification → structured, severity-graded findings + counts. Runs from the skill's bundled path (nothing is copied into the user's repo).SKILL.md) — the main agent drives the outer loop: call the workflow, apply fixes for confirmed critical/major/minor findings (nits opportunistically), commit oneAddress code review feedback (round N)per round, then re-review. Stops when the review is clean, on a no-progress guard, or after 8 rounds.Configurable model ladder (cost-efficient)
sonnet,opus, or an explicitmodels=…list.Reviewer agent
Prefers the
code-revieweragent (shipped by Claude Code's official review plugins) and falls back to the default workflow subagent when unavailable, so it runs anywhere.Registration
review-workflow/skill folder (SKILL.md,tile.json, bundled workflow)..claude-plugin/marketplace.json, the README skills table, and the tile-publish GitHub Action.