agentleFS
cloverinkcloverink/shipwright/skills/review-full/SKILL.md

review-full

Skill0 starsChanged 49 days agoMITView on GitHub

Smart review entry point. Auto-detects scope from the diff (frontend / backend / docs), spawns the right Opus reviewers in parallel, and keeps the same reviewers across fix rounds until the gates pass. --read-only for a quick scored peek with no edits. Use when unsure which review to run.

3 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 2

    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
  • could be automatic: a setting could enforce it
  1. Distinctive:

    All detected gates must pass.

    line 79
  2. Very specific:

    Never run /update-docs here.

    line 96
  3. Distinctive:

    Never push or merge.

    line 98

“Common” is measured: kept by at least 6 in 10 of the 3 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-full

Detects what changed, spawns the right reviewers in parallel, and runs the fix loops until the gates pass.

When to use: default entry point for any review. Not sure? Use this. Use a worker directly only if: you want to skip auto-detection. Called by: you, standalone. /ship runs the same lenses inside its A/R batch. Dispatches to: /review-code-fix · /review-ux-fix · /review-docs

Naming note: /review and /code-review are Claude Code built-ins. This skill's trigger is /review-full.

Modes

CommandModeWhat happens
/review-fullFix loopcode + UX loops (per detected scope), fixes applied
/review-full --read-onlyPeekspawn reviewers, report scores, change nothing
/review-full --code-onlyFix loopcode only, even on a frontend diff
/review-full --ux-onlyFix loopUX only

Step 1: Detect scope

BASE=$(git merge-base origin/main HEAD) || { echo "cannot resolve origin/main, git fetch first"; exit 1; }
CHANGED=$(git diff --name-only "$BASE"; git ls-files --others --exclude-standard)
<!-- CONFIGURE: replace globs with your project's paths -->
Frontend: src/components/**, src/pages/**, *.tsx, *.css
Backend:  src/api/**, src/services/**, *.service.ts, *.controller.ts
Docs:     docs/**, *.md
DetectedRuns
Frontend (with or without backend)code + UX
Backend onlycode
Docs only/review-docs
Nothing"no diff, nothing to review"

Step 2: Spawn reviewers in ONE tool-call block

Both reviewers read the same diff and neither writes, so they never wait for each other:

  • reviewer-code: code-reviewer (Opus), lens code, checklist <SKILLS_DIR>/review-code-fix/SKILL.md §Review focus
  • reviewer-ux: code-reviewer (Opus), lens ux, checklist <SKILLS_DIR>/review-ux-fix/SKILL.md §Review categories (frontend only; run the /review-ux-fix Step 0 precheck first)

SKILLS_DIR is the parent of this skill's base directory. Pass absolute paths: the agent has no Skill tool.

Each gets its own findings file (see Reviewer Continuity §Mechanics).

Step 3: Fix loop (skipped with --read-only)

This session runs both loops from /review-code-fix and /review-ux-fix:

  1. Union the findings, group by file, one writer per file, fix, lint.
  2. One commit per round covering both lenses, so the re-review can cite the SHA.
  3. Continue the same reviewers with SendMessage for rounds 2-3, in parallel. Round 4 = fresh escalation. A lens that already passed is not re-run unless the fixes touched its files.

Gates

ReviewGate
Code≥ 9.5 (9.0-9.4 fails)
UX= 10/10, zero open findings
Docsbinary pass/fail
<!-- CONFIGURE: thresholds -->

All detected gates must pass.

Report

## Review Report

Scope: frontend + backend · Mode: fix

| Review | Rounds | Score | Bar | Status |
|---|---|---|---|---|
| Code | 2 (continued) | 9.5 | ≥ 9.5 | ✅ |
| UX | 3 (continued) | 10 | = 10 | ✅ |

Guidelines

  • Never run /update-docs here. Docs are Phase D.
  • --read-only never commits, stages, or edits.
  • Never push or merge.

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.