codex-review
Skill15 starsChanged 8 months agoApache-2.0View on GitHub
Run OpenAI Codex CLI to review code changes made in the current session or branch. Invokes `codex review` with focused instructions derived from the actual changes. Use when the user says "codex review", "second opinion", "rival review", "cross-review", or "/codex-review".
91 other projects use a version of this.
Adopted by
Most recently taken up by
- p-chan/agent-skillstheir own version, first seen Nov 8, 2026
- p-chan/skillstheir own version, first seen Nov 8, 2026
- Smirnov-Labs/claude-skillstheir own version, first seen Nov 8, 2026
- juriwiens/skillstheir own version, first seen Nov 8, 2026
- ryanportfolio/winuse-mcptheir own version, first seen Nov 8, 2026
- fab-codes/agent-workflow-templatetheir own version, first seen Nov 8, 2026
Used with
Projects with it also use addskill, apply-firmware and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
ryanportfolioSkilladdskill
Create, import, update, or install repo or personal skills: runtime ownership, resources, discovery checks.
Also in 6 projects that use this
ryanportfolioSkillapply-firmware
Bring current folder (plain folder, existing repo, old or partial firmware copy) up to full Harness Firmware in place. Mirror external repo → adopt-repo.
Also in 6 projects that use this
- +10 more from ryanportfolio
Other versions91 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 91
- ClickHouse/ClickHouse★ 51ksame name
- chaseai-yt/claudex-loop★ 3.2ksame name
- morethanwords/tweb★ 2.7ksame name
- alinaqi/maggy★ 711same name
- alinaqi/claude-bootstrap★ 708same name
- ThibautBaissac/rails_ai_agents★ 665same name
- ai-analyst-lab/ai-analyst★ 304same name
- Sateezg/codex-bridge★ 248same name
- sd0xdev/sd0x-dev-flow★ 192same name
- sd0xdev/sd0x-harness★ 192same name
81 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository63 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- agent-browserSkill
- ast-grepSkill
- branch-review-before-prSkill
- browse-qaSkill
- browser-useSkill
- browseSkill
- bugfixSkill
- build-dmgSkill
- claude-reviewSkill
- codemapSkill
- code-simplifySkill
- commitSkill
- cost-estimateSkill
- create-prSkill
- find-agentsSkill
- find-bugsSkill
- frontend-design-ui-uxSkill
- git-merge-expertSkill
- git-merge-expert-worktreeSkill
- kiro-reviewSkill
- laravel-filamentSkill
- laravelSkill
- lokeiSkill
- map-project-monorepoSkill
- map-projectSkill
- nextjsSkill
- plan-founder-reviewSkill
- plan-to-task-list-with-dagSkill
- pr-retroSkill
- run-parallel-agents-feature-buildSkill
- run-parallel-agents-feature-debugSkill
- rustSkill
- secretsSkill
- startSkill
- update-agent-learningsSkill
- update-claude-learningsSkill
- update-claude-settingsSkill
- update-skill-learningsSkill
- ast-grepSkill
- branch-review-before-prSkill
- codemapSkill
- code-simplifySkill
- commitSkill
- cost-estimateSkill
- create-prSkill
- find-bugsSkill
- frontend-design-ui-uxSkill
- git-merge-expertSkill
- git-merge-expert-worktreeSkill
- map-project-monorepoSkill
- map-projectSkill
- plan-founder-reviewSkill
- plan-to-task-list-with-dagSkill
- pr-retroSkill
- run-parallel-agents-feature-buildSkill
- run-parallel-agents-feature-debugSkill
- startSkill
- update-agent-learningsSkill
- The first 60 shown.
The file
Show its rules and text
We split this skill into the rules it gives an agent
16 rules. 0 are common, 12 are distinctive, and 4 are very specific to this project.
0
Common
Most of the 79 edited versions we read keep these.
12
Distinctive
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 9
Use OpenAI Codex CLI to get an independent AI review of code changes.
- Distinctive: line 46
Read the diff to understand what areas need review focus.
- Very specific: line 57
Write a focused instruction file to /tmp/codex-review-instructions.md:
- Very specific: line 91
Always capture stderr with 2>&1 (codex logs to stderr)
- Distinctive: line 93
Run this in the background if the review is expected to take long:
- Very specific: line 94
Use run_in_background on the Bash tool
- Distinctive: line 99
Read the codex output and extract findings.
- Distinctive: line 114
If the user wants to fix the findings:
- Distinctive: line 115
Read each cited file to verify the finding is real
- Distinctive: line 117
Run tests and typecheck
- Distinctive: line 124
Update instructions to list all previously fixed issues
- Distinctive: line 125
Run round N+1 with the updated instructions
- Distinctive: line 128
Critical for iteration: Always add previously fixed issues to the instruction file's exclusion list.
- Very specific: line 188
Codex not installed: Check which codex first, tell user to install
- Distinctive: line 191
Repeated findings: Always maintain the exclusion list between rounds
- Distinctive: line 192
False positives: Always read the cited code before acting on findings
“Common” is measured: kept by at least 6 in 10 of the 79 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.
Files
Codex Review
Use OpenAI Codex CLI to get an independent AI review of code changes.
When to Use
- User explicitly asks for a codex review, second opinion, or cross-review
- User says "run codex", "codex review", "/codex-review"
- User wants an independent AI to verify Claude's work before merging
Prerequisites
codexCLI must be installed (which codex)- An OpenAI API key must be configured for codex
Workflow
Step 1: Determine the Review Scope
Identify what to review:
# Option A: Review current branch against main
git log --oneline main..HEAD | head -20
# Option B: Review uncommitted changes
git diff --stat
# Option C: Review a specific commit
git show --stat <sha>Pick the right codex review flag:
- Branch review:
--base main(or whatever the base branch is) - Uncommitted changes:
--uncommitted - Single commit:
--commit <sha>
Step 2: Analyze the Changes and Build Instructions
Read the diff to understand what areas need review focus. Build a focused instruction file based on the actual code changes.
# Get the diff summary
git diff --stat <base>..HEAD
# Identify key areas
git diff --name-only <base>..HEAD | sed 's|/[^/]*$||' | sort -uWrite a focused instruction file to /tmp/codex-review-instructions.md:
You are reviewing [brief description of what changed].
Focus your review on:
1. [Area 1 based on actual changes]
2. [Area 2 based on actual changes]
3. [Area 3 based on actual changes]
Verify each finding against the actual code. Read the relevant
source files before reporting.Key principles for instructions:
- Be specific about what changed — generic instructions get generic results
- List previously known issues so codex doesn't re-report them
- Focus on the areas where bugs are most likely (security, concurrency, edge cases)
- Tell codex to verify findings against actual code before reporting
Step 3: Run Codex Review
codex review \
--base <base-branch> \
--title "<descriptive title>" \
-c 'sandbox_permissions=["disk-full-read-access","disk-full-write-access","network-full-access"]' \
-c 'instructions="/tmp/codex-review-instructions.md"' \
2>&1Important flags:
sandbox_permissions— codex needs read access to verify findings against sourceinstructions— points to the focused review file from Step 2- Always capture stderr with
2>&1(codex logs to stderr)
Run this in the background if the review is expected to take long:
- Use
run_in_backgroundon the Bash tool - The output will be available when the command completes
Step 4: Parse and Report Findings
Read the codex output and extract findings. Look for lines matching:
[P1],[P2],[P3]— prioritized findingsFull review comments:— start of findings section- File paths with line numbers — specific locations
Report findings to the user in a clear table:
| # | Priority | Description | File:Line |
|---|----------|-------------|-----------|
| 1 | P1 | ... | ... |Step 5: Fix Findings (if requested)
If the user wants to fix the findings:
- Read each cited file to verify the finding is real
- Apply fixes
- Run tests and typecheck
- Commit the fixes
Step 6: Iterate (if requested)
For thorough review, run multiple rounds:
- Fix findings from round N
- Update instructions to list all previously fixed issues
- Run round N+1 with the updated instructions
- Repeat until findings converge to zero
Critical for iteration: Always add previously fixed issues to the instruction file's exclusion list. Otherwise codex will re-report the same issues every round, wasting time and API costs.
Example: Branch Review
# 1. Write instructions
cat > /tmp/codex-review-instructions.md << 'EOF'
You are reviewing a new authentication module.
Focus on:
1. SECURITY: JWT validation, token expiry, session management
2. CORRECTNESS: Error handling, edge cases, missing null checks
3. TESTING: Are there tests for error paths?
Verify each finding against the actual code before reporting.
EOF
# 2. Run review
codex review \
--base main \
--title "feat(auth): security review" \
-c 'sandbox_permissions=["disk-full-read-access"]' \
-c 'instructions="/tmp/codex-review-instructions.md"' \
2>&1Example: Iterative Deep Review
# Round 2 instructions (after fixing round 1 findings)
cat > /tmp/codex-review-instructions.md << 'EOF'
Round 2 review. Prior fixes (do NOT re-report):
- SQL injection in user query (parameterized)
- Missing auth check on /api/admin (added middleware)
- XSS in profile page (escaped output)
Focus on genuinely NEW issues only.
EOF
codex review --base main \
--title "feat(auth): review round 2" \
-c 'sandbox_permissions=["disk-full-read-access"]' \
-c 'instructions="/tmp/codex-review-instructions.md"' \
2>&1Safety Rules
| Rule | Reason |
| Never run codex without sandbox permissions | It needs file read access to verify findings |
| Always list prior fixes in iterative reviews | Prevents wasting rounds on already-fixed issues |
| Always verify findings before fixing | Codex can produce false positives |
| Run from the correct directory | Codex reviews the repo it's invoked in |
| Never pass secrets in instructions | The instruction file may be logged |
Failure Modes
- Codex not installed: Check
which codexfirst, tell user to install - No API key: Codex will fail with auth error — user needs to configure
- Too large diff: Codex may truncate; break into smaller review scopes
- Repeated findings: Always maintain the exclusion list between rounds
- False positives: Always read the cited code before acting on findings
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
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.

