Install the "codex-review-loop" agent skill from https://github.com/aafqaq/codex-lb-enhanced/tree/main/.agents/skills/codex-review-loop into .claude/skills/codex-review-loop/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "codex-review-loop", then confirm the skill loads.
Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Type this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
skills CLI
$ npx skills add aafqaq/codex-lb-enhanced --skill codex-review-loop -a codex
Project install goes to .agents/skills/; add -g for ~/.codex/skills/.
Install the "codex-review-loop" agent skill from https://github.com/aafqaq/codex-lb-enhanced/tree/main/.agents/skills/codex-review-loop into .agents/skills/codex-review-loop/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "codex-review-loop", then confirm the skill loads.
Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
skills CLI
$ npx skills add aafqaq/codex-lb-enhanced --skill codex-review-loop -a cursor
Project install goes to .agents/skills/; add -g for ~/.cursor/skills/.
Install the "codex-review-loop" agent skill from https://github.com/aafqaq/codex-lb-enhanced/tree/main/.agents/skills/codex-review-loop into .cursor/skills/codex-review-loop/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "codex-review-loop", then confirm the skill loads.
Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
skills CLI
$ npx skills add aafqaq/codex-lb-enhanced --skill codex-review-loop -a gemini-cli
Project install goes to .agents/skills/; add -g for ~/.gemini/skills/.
Install the "codex-review-loop" agent skill from https://github.com/aafqaq/codex-lb-enhanced/tree/main/.agents/skills/codex-review-loop into .gemini/skills/codex-review-loop/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "codex-review-loop", then confirm the skill loads.
Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Installs for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
skills CLI
$ npx skills add aafqaq/codex-lb-enhanced --skill codex-review-loop -a github-copilot
Project install goes to .agents/skills/; add -g for ~/.copilot/skills/.
Install the "codex-review-loop" agent skill from https://github.com/aafqaq/codex-lb-enhanced/tree/main/.agents/skills/codex-review-loop into .github/skills/codex-review-loop/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "codex-review-loop", then confirm the skill loads.
GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
skills CLI
$ npx skills add aafqaq/codex-lb-enhanced --skill codex-review-loop -a opencode
OpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
Install the "codex-review-loop" agent skill from https://github.com/aafqaq/codex-lb-enhanced/tree/main/.agents/skills/codex-review-loop into .opencode/skills/codex-review-loop/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "codex-review-loop", then confirm the skill loads.
OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
Skill name
codex-review-loop
GitHub stars
102
Token cost
~1.7k tokens
SKILL.md length
824 words
Files
4 (incl. scripts, references)
Skills in repo
4
Repo updated
First seen
Licence
MIT
At a glance
Adversarial PR code review using Codex CLI. An agent skill from aafqaq/codex-lb-enhanced.
Works in 5 steps: Scope Resolution → Codex Review → Finding Analysis → …
DevOps & Cloud work in your project
SKILL.md covers Arguments, Phases, Phase 1: Scope Resolution and Phase 2: Codex Review, plus 5 more sections
Runs Shell scripts from its folder; calls bash, gh and codex
What it does
Codex Review Loop is an agent skill from aafqaq/codex-lb-enhanced. Adversarial PR code review using Codex CLI. Codex review (~20-50 min) - structured findings - HITL approval - fix loop.
Its SKILL.md is about 1.7k tokens, which your agent loads only when the skill is triggered. The skill folder holds 7 other files, including scripts and reference files (for example `references/prompts/adversarial-review.md`, `references/schemas/review-findings.md` and `scripts/codex-subagent.sh`).
It sits in DevOps & Cloud. The repository describes itself as: Production-ready Codex-compatible account-pool gateway with seamless failover, full-session recovery, native quota headers, WebSocket/HTTP bridge, and observability. The licence is MIT.
When your agent uses it
DevOps & Cloud work in your project
Example prompts
“/codex-review-loop”
Requirements
Node.js
A Bash shell
Workflow steps
5 steps, taken from the step headings in SKILL.md.
Read from SKILL.md and the folder at commit c0e9e18. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Ships 1 file in scripts/ (Shell), which the agent can run.
Shell commands in SKILL.md call:
bash
gh
codex
uvx
uv
npm
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md. Its commands use gh, uvx, uv and npm, which can reach the network depending on how they are called.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
Codex Review Loop loads about 1.7k tokens when it runs, and up to ~4.1k if it reads all its reference files. Until then it costs about 35 tokens; SKILL.md has 824 words of instructions outside code blocks.
Always· name and description, kept in context so the agent knows when to use it
~35
When it runs· the whole SKILL.md, loaded when a task matches
~1.7k
With references· SKILL.md plus every file in references/, read only if the agent opens them
~4.1k
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.
Download SKILL.mdSave it as .claude/skills/codex-review-loop/SKILL.md (or your agent's skills folder). This skill also uses 3 other files; get the full folder from GitHub.
Note: Codex CLI v0.105.0+ does not support combining --base/--commit
with a custom prompt. The built-in review logic is used automatically.
For --uncommitted mode (no diff target), pipe stdin for custom instructions:
Inform the user the review is running (~20-50 min).
The script parses Codex output and returns the final review text. Review
rollouts intentionally remain persistent. If terminal output is lost, use
codex resume --include-non-interactive to locate the review, or
codex resume <SESSION_ID> when its ID is known. Do not add
--ephemeral to the wrapper. The wrapper also relies on the configured
non-interactive approval/sandbox policy because Codex CLI 0.147.0 removed
the historical --full-auto argument from exec review.
Error handling
Exit code
Meaning
Action
0
Success
Proceed to Phase 3
1
Codex error
Show error, offer retry or abort
127
codex not found
Guide: npm i -g @openai/codex
Environment overrides
Variable
Purpose
CODEX_REVIEW_MODEL
Override Codex model
CODEX_REVIEW_REASONING
Override reasoning effort
Phase 3: Finding Analysis
Parse the raw Codex output into structured findings.
References
Schema: references/schemas/review-findings.md — field definitions,
severity/category/effort enums, status lifecycle
Convention rules: .agents/skills/project-conventions/conventions.md —
project coding conventions to cross-reference
Severity escalation/downgrade guide
Escalate to Critical: Unvalidated external input, secret exposure,
injection vectors, data integrity compromise
Escalate to High: API contract change without test update, possible
NoneType error, blocking I/O in async context
Downgrade to Medium: Correct functionality but violates conventions,
duplicate logic, unnecessary complexity
Downgrade to Low: Pure style, unused import, missing docstring
Structure violations (core/modules boundary, DI bypass) -> architecture
Testing gaps (contract change without test update) -> testing
Process
Parse each finding from the raw output
Assign severity and category using the schema definitions and guide above
Cross-reference with project conventions
Estimate fix effort
Deduplicate (keep the most severe instance)
Sort: severity desc, file path asc
Show full SKILL.md (331 more words)Show less
HITL Gate 1: Finding Review
Present all findings to the user as a summary table
(ID, severity, category, file:line, title, effort).
Ask the user to choose a fix scope:
Fix all (recommended)
Fix Critical/High only
Report only (no fixes)
Phase 4: Atomic Fix Loop
For each approved finding, execute an atomic fix-verify-commit cycle.
Per finding:
Fix: Read target file(s), apply the fix.
HITL Gate 2 (conditional): For Critical/High findings that modify existing
logic (not just adding new code), show the current and proposed code to the
user for confirmation. Medium/Low findings are auto-fixed.
Verify (all must pass):
uvx ruff check .
uvx ruff format --check . (auto-fix and re-check if needed)
uv run ty check
uv run pytest (mapped test files, or full suite if no mapping found)
Commit: fix(review): P{i} - {title}
On failure: Roll back changes, mark finding as skipped, continue to next.
Phase 5: Re-review Loop
After all fixes are committed, automatically re-run the Codex review
to check for regressions or new issues introduced by the fixes.
Loop behaviour
Re-run Codex review (--base <branch>) against the same base.
Parse findings (Phase 3) and present to user (HITL Gate 1).
If 0 findings → loop terminates, proceed to Final Report.
If findings exist → execute Phase 4 (fix-verify-commit), then repeat
from step 1.
Safety limits
Max iterations: 3 (initial review + 2 re-reviews).
After 3 iterations, terminate the loop regardless of remaining findings
and output the Final Report with unresolved items noted.
Escalation: If the same finding recurs across 2 consecutive iterations,
mark it as wont_fix and skip in subsequent iterations.
HITL override
At each re-review result, the user may choose:
Continue — proceed with fixes (default when findings > 0)
Stop — terminate the loop early and output Final Report
Final Report
Output a summary including:
PR metadata (number, title, base/head)
Loop iteration count and termination reason (clean / max iterations / user stop)
Codex Review Loop next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
Codex Review Loop compared with similar skills
Skill
Stars
Used in
Tokens
Auto-check
Licence
Repo updated
Codex Review Loop this skillaafqaq/codex-lb-enhanced
A skill your agent uses when the user asks how to build with OpenAI products or APIs and needs up-to-date official documentation with citations (for example: Codex, Responses API, Chat Completions…
OpenSpec context documentation policy: keep requirements in spec.md and capture narrative context in context/overview/rationale docs under openspec/specs.
Adversarial PR code review using Codex CLI. An agent skill from aafqaq/codex-lb-enhanced. Codex Review Loop is an agent skill from aafqaq/codex-lb-enhanced. Adversarial PR code review using Codex CLI.
When should I use Codex Review Loop?
Codex Review Loop fits situations like: devOps & Cloud work in your project.
How do I install Codex Review Loop in Claude Code?
Run `npx skills add aafqaq/codex-lb-enhanced --skill codex-review-loop -a claude-code`. Or copy the skill folder (.agents/skills/codex-review-loop in aafqaq/codex-lb-enhanced) into .claude/skills/codex-review-loop in your project. Claude Code loads it when a task matches its description.
How do I install Codex Review Loop in Codex?
Run `npx skills add aafqaq/codex-lb-enhanced --skill codex-review-loop -a codex`. Or copy the skill folder (.agents/skills/codex-review-loop in aafqaq/codex-lb-enhanced) into .agents/skills/codex-review-loop in your project. Codex loads it when a task matches its description.
Can I use Codex Review Loop in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add aafqaq/codex-lb-enhanced --skill codex-review-loop -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/codex-review-loop, .gemini/skills/codex-review-loop, .github/skills/codex-review-loop and .opencode/skills/codex-review-loop in your project.
What does Codex Review Loop need to run?
Going by SKILL.md and its folder, Codex Review Loop needs a shell for the scripts in its folder and the command-line tools its instructions call (bash, gh, codex, uvx, uv and npm). Our summary lists: Node.js; A Bash shell.
Does Codex Review Loop access the network?
SKILL.md contains no URLs. Its commands use gh, uvx, uv and npm, which can reach the network depending on how they are called. This is read from the text; nothing was executed.
Is Codex Review Loop safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.
What licence does Codex Review Loop use?
Codex Review Loop is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Codex Review Loop use?
About 1.7k tokens (SKILL.md is roughly 6.8k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 2.4k tokens, read only when the agent opens those files.
What are the alternatives to Codex Review Loop?
Skills that share tags, products or a category with Codex Review Loop: Debug (sbusso/claudeclaw, 194 stars), AI Server (Opentrons/opentrons, 521 stars), Kopiur Design (home-operations/kopiur, 113 stars) and GitHub Actions Pat (bifrost-proxy/bifrost, 160 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Codex Review Loop?
aafqaq (a GitHub user) maintains it in aafqaq/codex-lb-enhanced, which has 102 GitHub stars. The repository holds 4 skills in this directory. The repository was last updated on October 5, 2026.
Source: aafqaq/codex-lb-enhanced on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.
