From codex-exec
Delegate code review, plan review, and exploration to Codex CLI. Use when you need an independent second opinion on code changes, plans, or architecture — or when asked to run codex review/exec.
How this skill is triggered — by the user, by Claude, or both
Slash command
/codex-exec:codex-execThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Delegate review and exploration tasks to **Codex CLI** for an independent second opinion. Codex acts as a strict counter-reviewer that catches blind spots you might miss.
Delegate review and exploration tasks to Codex CLI for an independent second opinion. Codex acts as a strict counter-reviewer that catches blind spots you might miss.
Use this skill when:
codex exec, codex review, or to "get codex's opinion"Do NOT use this skill for tasks where you are confident in the output and the user hasn't asked for a second opinion. Reserve it for quality gates and deliberate review steps.
Review the current branch's changes against a base branch:
# Review current branch against main (most common)
codex review --base main
# Review only uncommitted changes (staged + unstaged + untracked)
codex review --uncommitted
# Review a specific commit
codex review --commit <sha>
# Review with custom focus instructions
codex review --base main 'Pay special attention to error handling and edge cases'
# Review with a descriptive title for context
codex review --base main --title "Add user authentication middleware"
Use codex exec for reviews that go beyond standard code review — plan reviews, architecture critique, or targeted analysis:
# Review a plan or design document
codex exec 'Review the plan in PLAN.md. Be a strict critic: identify gaps, missing edge cases, and over-engineering. Suggest concrete improvements.'
# Targeted code analysis
codex exec 'Look at the authentication middleware in src/auth/. Are there any security concerns? What about rate limiting and session handling?'
# Deep exploration for blind spots
codex exec 'Explore the codebase for inconsistencies, dead code, and simplification opportunities. Focus on the src/api/ directory.'
# Review specific files
codex exec 'Review src/handlers/webhook.ts for correctness, error handling, and clarity. Be pointed in your response — list problems with file:line references.'
Run review in a loop until all issues are resolved:
# First pass
codex review --base main
# Fix the issues codex found, then re-run
codex review --base main
# Repeat until the review comes back clean
For plans, iterate with codex exec:
codex exec 'Review the plan in PLAN.md and list remaining problems.'
# Revise the plan based on feedback, then re-run until agreed
After completing implementation and before pushing:
codex review --base mainWhen drafting an implementation plan:
codex exec 'Review the plan in PLAN.md. Be very strict: identify gaps, missing edge cases, wrong assumptions, and over-engineering.'When exploring a codebase area for quality improvements:
# Broad exploration
codex exec 'Do a deep dig of src/. Find blind spots, interesting things to fix, and things to simplify. Be specific with file:line references.'
# Follow up on specific findings
codex exec 'Look deeper at the error handling pattern in src/api/client.ts. Is the retry logic correct? What happens on timeout?'
--base main for branch reviews, --uncommitted for work-in-progress checks.-m <model> to pick a specific model for the review if needed: codex review --base main -m o3.npx claudepluginhub mrshu/agent-skills --plugin codex-execRuns cross-model code reviews using the external Codex CLI tool from a Claude session. Catches bugs that single-model self-review would miss by leveraging a different reviewer architecture.
Cross-model review using OpenAI Codex to independently verify plans or code diffs, iterating up to 5 rounds. Useful for architecture decisions, non-trivial refactors, and critical config changes.
Orchestrates cross-agent code review: Claude implements plans/code, Codex CLI reviews via bash scripts. Activates on 'codex review' or Russian triggers.