codex-review

name	codex-review
description	Requests an independent code review from OpenAI Codex CLI, critically evaluates its findings, and applies warranted fixes. Activates when: the user says /codex-review, asks for a Codex review, or wants an external AI review of changes.
user_invocable	true

Codex Code Review

Run an independent code review using OpenAI Codex, then critically evaluate and apply warranted findings.

Prerequisites

Both the codex-plugin-cc and the global Codex CLI must be installed.

Plugin — if /codex:review is not available:

/plugin marketplace add openai/codex-plugin-cc
/plugin install codex@openai-codex
/reload-plugins
/codex:setup

Codex CLI — if the companion script reports "Codex CLI is not installed":

npm install -g @openai/codex

Auth: !codex login if not authenticated.

Step 1: Determine what to review

Check what has changed:

git diff --stat HEAD
git diff --stat --staged

Decide the review scope:

• Uncommitted changes — use --scope working-tree
• Feature branch vs master — use --base master

Step 2: Run Codex review

Find the companion script:

COMPANION=$(ls ~/.claude/plugins/cache/openai-codex/codex/*/scripts/codex-companion.mjs 2>/dev/null | sort -V | tail -1)

For uncommitted changes:

node "$COMPANION" review --scope working-tree --background

For uncommitted changes with focus:

FOCUS="focus on Blade directive output safety and Octane request isolation"
node "$COMPANION" adversarial-review --scope working-tree --background "$FOCUS"

For feature branch vs master:

node "$COMPANION" review --base master --background

For feature branch with focus:

FOCUS="<user argument>"
node "$COMPANION" adversarial-review --base master --background "$FOCUS"

Poll for completion:

TIMED_OUT=true
for i in $(seq 1 15); do
  sleep 20
  STATUS=$(node "$COMPANION" status 2>&1)
  if ! echo "$STATUS" | grep -qE "\| running \||\| queued \|"; then
    TIMED_OUT=false
    break
  fi
  echo "Still running... ($i)"
done

if [ "$TIMED_OUT" = "true" ]; then
  echo "Timed out — check manually: node \"$COMPANION\" status"
else
  node "$COMPANION" result
fi

The result command without a job ID returns the latest finished job.

Step 3: Critically evaluate findings

Codex findings are suggestions, not mandates. For each finding:

1. Is it a real bug? — Verify by reading the code. Don't trust Codex's assessment blindly.
2. Is it already tested? — Check if existing tests cover the scenario.
3. Is it a style preference? — Skip. Don't change working code for style.
4. Is it a false positive? — Codex may misunderstand Laravel internals, Testbench setup, or the package's architecture. Verify against the actual behavior.

Step 4: Apply warranted fixes

For findings that are genuine issues:

1. Fix the code
2. Run vendor/bin/phpunit to verify tests
3. Run vendor/bin/phpstan analyse --memory-limit=2G to verify static analysis
4. Run vendor/bin/pint --dirty --format agent to verify style

Step 5: Report

Summarize to the user:

## Codex Review Summary

### Applied
- [Issue] — [What was wrong and how you fixed it]

### Dismissed
- [Finding] — [Why it was dismissed: false positive / already tested / style preference]

### No Issues
- [Categories that were clean]