// 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.
Profile a slow request, command, or code path with sandermuller/stopwatch. Activate when the user mentions: slow request, slow endpoint, slow page, slow query, slow job, slow command, slow API call, slow HTTP, performance issue, profile this, why is this slow, where is the time going, optimize, bottleneck, latency, n+1, query count, memory usage, http calls, outbound requests, server-timing.
Pre-push / pre-release checklist. Runs Rector, Pint, Pest, PHPStan, audits README + `.ai/` docs for staleness, then commits + pushes, watches CI for green, and drafts release notes against the verified SHA. Activate before: pushing to remote, tagging a release, writing release notes, or when user mentions: pre-release, pre-push, release checklist, ship, cut release, release notes.
MUST USE when creating, updating, or managing AI skills and guidelines. Activates when: creating a new skill, editing existing skills, updating coding guidelines, modifying AI instructions, working with .ai/ directory, or when user mentions: skill, CLAUDE.md, guidelines, AI configuration.
Runs backend code quality checks in two tiers: Pint + related tests (every change), PHPStan + full suite (completion only). Activate after making changes to PHP files, or when user mentions: phpstan, pint, code quality, static analysis, code style, run checks.
Implements a specification file phase-by-phase with progress tracking. Activates when: implementing a spec, building from a spec, starting a spec phase, or when user mentions: implement spec, spec file, implement phase, build spec, start phase.
Writes implementation-ready specification files with progress-trackable phases. Activates when: writing a spec, creating a spec file, documenting a feature plan, or when user mentions: write spec, create spec, spec format, spec template.
| 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 |
Run an independent code review using OpenAI Codex CLI, then critically evaluate and apply warranted findings.
Check what has changed:
git diff --stat HEAD
git diff --stat --staged
If there are uncommitted changes, review those (--uncommitted). If the working tree is clean, review the latest commit (--commit HEAD).
Run the appropriate command:
For uncommitted changes:
codex exec review --full-auto --uncommitted "Review for correctness, security, edge cases, and test coverage gaps. This is a Laravel validation package. Focus on: rule compilation ordering, fast-check correctness, FormRequest lifecycle compatibility, and cross-field wildcard handling. Be concise — only report real issues, not style preferences."
For the latest commit:
codex exec review --full-auto --commit HEAD "Review for correctness, security, edge cases, and test coverage gaps. This is a Laravel validation package. Focus on: rule compilation ordering, fast-check correctness, FormRequest lifecycle compatibility, and cross-field wildcard handling. Be concise — only report real issues, not style preferences."
For changes against main:
codex exec review --full-auto --base main "Review for correctness, security, edge cases, and test coverage gaps. This is a Laravel validation package. Focus on: rule compilation ordering, fast-check correctness, FormRequest lifecycle compatibility, and cross-field wildcard handling. Be concise — only report real issues, not style preferences."
Codex findings are suggestions, not mandates. For each finding:
For findings that are genuine issues:
vendor/bin/pest --no-coverage to verifyvendor/bin/phpstan analyse src/ --memory-limit=2G to verifySummarize 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]