guide

A guide to the Codex version of the review-loop workflow, which coordinates planning, implementation, and review. It uses shared configuration and session logs so Codex and Claude Code can work from the same state.

In plain words
What is it for?
Use it to understand the review-loop lifecycle, shared files, reviewer choices, and what Codex Stage 1 currently supports.
Why use it?
It explains how the workflow is organized and where its current boundaries are. Shared files reduce confusion when work moves between the two coding tools.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/nytc69/review-loop/guide
Any agent
npx skills add NYTC69/review-loop --skill guide
Clone the repo
git clone --depth 1 https://github.com/NYTC69/review-loop

Made for: Claude Code, Codex.

Per session 21 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 589 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00021 $0.00589
Opus 5 $0.00010 $0.00295
Sonnet 5 $0.00004 $0.00118
Haiku 4.5 $0.00002 $0.00059

Measured 2d ago against content hash f7ca3330da84, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

guide scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.agents/skills/guide/SKILL.md · 68 lines

How it starts

The opening of the file, as written. The whole thing — 68 lines — stays where its author put it; the contents beside it link to each section on GitHub.

review-loop Guide

What review-loop Does in Codex

In Codex Stage 1, review-loop is a repo skill that follows the same broad review workflow used in Claude Code. It coordinates planning, implementation, and review while keeping the shared session log current. Codex Stage 1 follows the same broad exec -> polish -> docs -> security -> delivery lifecycle.

Codex reads and writes the same review-loop state as Claude Code:

  • .review-loop/config.md
  • .review-loop/sessions/

That means a project can keep one shared config file and one shared session log history across both runtimes. Codex Stage 1 assumes a single orchestrator-owned workspace for the session.

Stage 1 Scope

Stage 1 in Codex includes only:

  • review-loop
  • guide

It does not yet migrate:

  • code-quality-loop
  • review-pr
  • reorganize

Reviewer Behavior

Codex Stage 1 defaults to the outside-sandbox Claude CLI reviewer path. In practice, that means review stays on claude -p --model ... unless the user explicitly opts into the local Codex reviewer.

You can force the local Codex reviewer with:

  • codex_reviewer_backend: codex

This is the override to use when you want Codex to skip the Claude CLI reviewer and use the Codex reviewer directly. In that case, codex_reviewer_model is the paired model override, while reviewer_model still applies to the Claude CLI reviewer path and judgment_model is its shared-tier fallback before the explicit claude-sonnet-4-6 backstop.

cheap_model is accepted in the shared config so Claude and Codex can share the same file, but in Codex Stage 1 it is a documented no-op because only judgment-tier Codex agents are currently shipped. quality_focus applies only when Step 3.5 Quality Polish actually runs. skip_quality_polish: true mints polish as a no-op completion and still continues through docs and security.

Usage Notes

  • Executor-created hidden worktrees are forbidden in Codex Stage 1.
  • Codex Stage 1 supports before-polish, before-docs, and before-security as clean stop points.
  • Codex repo skills live under .agents/skills/ in the Codex workspace.
  • Keep the shared review-loop config in .review-loop/config.md.
  • Keep session logs in .review-loop/sessions/.
  • Use the local Codex reviewer only when you need to bypass the default Claude CLI reviewer path explicitly.

Read the full file on GitHub · 68 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 68 lines · 21 tokens per session scan A f7ca3330da84

Subscribe to this mod's changes

guide is a skill published in the GitHub repository NYTC69/review-loop (3 stars, last pushed 3mo ago), licensed Apache-2.0. It adds 21 tokens to every session and 589 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

codex-adversarial-review-lite

Codex Adversarial Review - Lite: user-invoked audit workflow for Claude Code users who want Codex CLI to independently review AI-generated code, plans, test expectations, and scope before fixes are applied. Cross-platform (Windows, macOS, Linux, WSL). Use only when the user explicitly invokes audit or selftest.

razaumair2203-ux/codex-adversarial-review-lite · 76 tokens

codex-handoff

Three-phase collaboration workflow between Claude Code (planner/decision-maker) and Codex CLI (implementer/reviewer) via the codex-plugin-cc plugin. Use whenever a coding task involves multiple files, new modules, cross-cutting refactors, business logic changes, or anything that warrants a written spec and a review…

ParaGenie/claude-codex-handoff · 175 tokens

brain-docs

Set up and maintain a "project brain" — a brain/ folder holding three living documents (decisions.md, flow.md, todo.md) that give a project persistent memory across chat sessions. Use this skill whenever the user wants a decision log, architecture/flow notes, project memory, living documentation, or notes that survive…

IAbhishekJhaI/brain · 162 tokens

codex-handoff

Three-phase collaboration workflow between Claude Code (planner/decision-maker) and Codex CLI (implementer/reviewer) via the codex-plugin-cc plugin. Use whenever a coding task involves multiple files, new modules, cross-cutting refactors, business logic changes, or anything that warrants a written spec and a review…

ParaGenie/claude-codex-handoff · 175 tokens

moai-ref-ui-polish

UI polish and interface-completion reference: the small visual details — concentric border radius, optical alignment, shadow-vs-border, motion easing, typography smoothing, tabular numbers, icon stroke weight, hit areas — that separate polished interfaces from generic ones. Agent-extending skill that amplifies…

modu-ai/moai-adk · 98 tokens

trellis-brainstorm

Guides collaborative requirements discovery before implementation. Creates task directory, seeds PRD, asks high-value questions one at a time, researches technical choices, and converges on MVP scope. Use when requirements are unclear, there are multiple valid approaches, or the user describes a new feature or complex…

mindfold-ai/Trellis · 65 tokens