What is codex-review?

A review tool that sends an implementation plan to Codex, another coding assistant, and returns written feedback with an approval or changes-needed decision.

Use it in plan debate rounds to review plan files, save the feedback, and track whether the plan was approved.

It adds a separate review of a plan before implementation, making missing or unclear work easier to find.

codex-review — Skill for Claude Code and Codex by closedloop-ai

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.

Any agent

npx skills add closedloop-ai/claude-plugins --skill codex-review

Clone the repo

git clone --depth 1 https://github.com/closedloop-ai/claude-plugins

Made for: Claude Code, Codex.

Or install code, the plugin that ships this one along with the rest of its 15 skills, 7 commands, 23 agents, 7 hooks.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/skills/closedloop-ai/claude-plugins/codex-review.svg)](https://agentmods.dev/skills/closedloop-ai/claude-plugins/codex-review)

Your own site

<a href="https://agentmods.dev/skills/closedloop-ai/claude-plugins/codex-review"><img src="https://agentmods.dev/badge/skills/closedloop-ai/claude-plugins/codex-review.svg" alt="Measured on agentmods" height="20"></a>

Per session 51 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 1,409 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

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.

Model	Per session	Once invoked
Fable 5.1	$0.00051	$0.01409
Opus 5	$0.00026	$0.00705
Sonnet 5	$0.00010	$0.00282
Haiku 4.5	$0.00005	$0.00141

Measured yesterday against content hash ff9104ee01dd, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade A, and why

codex-review 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 yesterday.

The scan reads SKILL.md. This mod also ships 1 executable file (scripts/run_codex_review.sh), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

plugins/code/skills/codex-review/SKILL.md · 97 lines

How it starts

The opening of the file, as written. The whole thing — 97 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Codex Review

Call Codex to review an implementation plan and return structured feedback with an approval verdict.

When to Use

Activated once per debate round in the /plan-with-codex command, before Claude revision. The orchestrator calls this skill to get Codex's assessment of the current plan.

Usage

bash ${CLAUDE_SKILL_DIR}/scripts/run_codex_review.sh \
  --plan-file <path> \
  --feedback-file <path> \
  --request-file <path> \
  --revisions-file <path> \
  --round <N> \
  --codex-model <model> \
  [--session-id <thread_id>] \
  [--log-id <uuid>]

Argument	Required	Default	Description
`--plan-file`	Yes	--	Absolute path to the implementation plan (plan.json or plan.md) that Codex will review. The script injects this path into the review prompt so Codex can read and analyze the plan's contents.
`--feedback-file`	Yes	--	Absolute path where the script writes Codex's full feedback text (parsed from the JSON stream). The orchestrator reads this file after the script completes to get the detailed findings. This file is overwritten each round.
`--request-file`	No	--	Absolute path to the original user request sidecar. When present and non-empty, the script tells Codex to read it before reviewing the plan so it can judge whether the plan chose the right overall approach for the request, not just whether the plan is internally consistent. If the file begins with `[synthesized]`, Codex is told to treat it as a weak hint rather than authoritative user intent.
`--revisions-file`	No	--	Absolute path to Claude's revision summary from the previous round, listing which findings were accepted and which were rejected with evidence. Only meaningful when round > 1 AND the file exists with actual content (the script checks `-s` for non-empty). The script injects this path into Codex's prompt so it can read the revisions before re-reviewing, but it explicitly tells Codex to verify Claude's rebuttals against the updated plan and codebase rather than trusting the summary blindly. Omit entirely on round 1 or when no revisions file has been written yet -- do not pass `/dev/null` or an empty file.
`--round`	No	1	The current debate round number (1-indexed). Controls the review prompt phase: round 1 runs a broad but material audit, rounds 2-4 run a delta review that first checks whether prior findings were resolved, and rounds 5+ run a blocker-only convergence review. Also gates whether the revisions file is included in the prompt.
`--codex-model`	No	gpt-5.3-codex	The OpenAI model ID passed to `codex -m`. Controls which model performs the review.
`--session-id`	No	--	Codex thread ID returned as `CODEX_SESSION` from a previous round. When provided, the script attempts `codex exec resume <session_id>` to continue the conversation with full prior context. If resume fails, it falls back to a fresh session automatically. Omit on round 1.
`--log-id`	No	auto-generated UUID	Identifier for the persistent JSONL log file at `~/.closedloop-ai/plan-with-codex/<log-id>.jsonl`. The raw Codex JSON stream is appended here each round. Pass the same ID across all rounds of a debate to keep the full conversation history in one file. If omitted, a new UUID is generated.

Read the full file on GitHub · 97 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

scripts/run_codex_review.sh 16 KB runs code

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.

yesterday First seen · 97 lines · 51 tokens per session scan A ff9104ee01dd

Subscribe to this mod's changes

codex-review is a skill published in the GitHub repository closedloop-ai/claude-plugins (103 stars, last pushed 2d ago), licensed Apache-2.0. It adds 51 tokens to every session and 1,409 once invoked, about $0.0003 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 2026-09-07.

Code review LLM Planning agentic-ai ai-agents anthropic developer-tools

pr-reviewer

Reviews a diff or security scope read-only using evidence-tiered findings, structural and context-error rubrics, and repository review policy. Use when asked to "review my changes", "structural review", "review for AI patterns", or "security audit". For applying fixes use tidy; for UI defects use ui-design.

mblode/agent-skills · 68 tokens

pr-babysitter

Monitors or repairs an open GitHub PR: CI failures, conflicts, review threads, and merge readiness, reporting state changes. Use when asked to "watch this PR", "fix CI", "resolve conflicts", or "address review comments". For PR metadata use pr-creator; for npm release PRs use autoship.

mblode/agent-skills · 72 tokens

dx-audit

Audits libraries, CLIs, and SDKs using 38 rules for public contracts, package exports, piped output, errors, and configuration. Use when asked to "audit my CLI", "review my SDK", "make this agent-friendly", or diagnose package type resolution. For agentic product trust use ax-audit; for docs use docs-writing.

mblode/agent-skills · 76 tokens

pr-creator

Creates or updates GitHub PRs with house-style titles and bodies, issue linking, repository templates, and draft state. Use when asked to "create a PR", "rewrite the PR description", or "polish this PR". For code fixes use tidy; for CI and review threads use pr-babysitter; for npm releases use autoship.

mblode/agent-skills · 75 tokens

tidy

Applies diff-scoped simplifications using reuse, ownership, efficiency, and test-value checks, including actionable review findings. Use when asked to "tidy this", "simplify my diff", or "apply the review findings". For a read-only report use pr-reviewer; for repository architecture use codebase-architecture.