collaborating-with-codex

collaborating-with-codex is a skill for Claude Code, Codex from briandconnelly/codex-in-claude. It costs 141 tokens per session (3,989 once invoked), scanned A, original, MIT.

A workflow guide for having Claude Code work with OpenAI Codex. It covers consultations, delegated coding, independent reviews, and second opinions.

In plain words
What is it for?
Use it when Claude Code needs Codex to review code, implement a task in an isolated Git worktree, provide another opinion, or run a structured review-and-revise process.
Why use it?
It defines checks and safety rules for deciding when Codex can be called and how its results should be handled.

Skill for Claude CodeCodex

Written for Claude Code and Codex: shipped in a Claude Code plugin, but also reads ~/.codex or $CODEX_HOME. Also seen: mentions Claude Code; mentions AGENTS.md; mentions Codex.

Part of the codex-in-claude plugin — 8 skills, 1 MCP server shipped together

Good fit Use it when Claude Code needs Codex to review code, implement a task in an isolated Git worktree, provide another opinion, or run a structured review-and-revise process.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/briandconnelly/codex-in-claude/collaborating-with-codex
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 briandconnelly/codex-in-claude --skill collaborating-with-codex
Clone the repo
git clone --depth 1 https://github.com/briandconnelly/codex-in-claude

Made for: Claude Code, Codex.

Or install codex-in-claude, the plugin that ships this one along with the rest of its 8 skills, 1 MCP server.

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.

agentmods badge for collaborating-with-codex

README.md
[![agentmods](https://agentmods.dev/badge/skills/briandconnelly/codex-in-claude/collaborating-with-codex/github.svg)](https://agentmods.dev/skills/briandconnelly/codex-in-claude/collaborating-with-codex)
Your own site
<a href="https://agentmods.dev/skills/briandconnelly/codex-in-claude/collaborating-with-codex"><img src="https://agentmods.dev/badge/skills/briandconnelly/codex-in-claude/collaborating-with-codex/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for collaborating-with-codex

Your own site · 80×15
<a href="https://agentmods.dev/skills/briandconnelly/codex-in-claude/collaborating-with-codex"><img src="https://agentmods.dev/badge/skills/briandconnelly/codex-in-claude/collaborating-with-codex.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 141 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,989 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. 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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00141 $0.03989
Opus 5 $0.00071 $0.01995
Sonnet 5 $0.00028 $0.00798
Haiku 4.5 $0.00014 $0.00399

Measured 9d ago against content hash e37133130950, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

collaborating-with-codex scanned grade A with 1 finding 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 9d 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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

operations, `gh`, `curl`, or publishing inside it, because delegate runs have no network egress
skills/collaborating-with-codex/SKILL.md · 228 lines

How it starts

The opening of the file, as written. The whole thing — 228 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Collaborating with Codex

Use this skill as the router and shared safety contract for every Codex workflow. Retain responsibility for the work, and compose this guidance with applicable planning, debugging, review, and verification skills instead of replacing them.

Shared workflow

  1. Call codex_status before a paid call. Proceed only when both ready: true and extra_args_valid: true. If either is false, stop and surface the corresponding readiness or operator-configuration detail.
  2. Treat rate_limit as advisory, with one exception. codex_status reads it live from the Codex app-server (no model spend, nothing persisted), so it is current when ready: true. Decide spend from it: proceed on available; defer non-urgent calls on limited or exhausted; treat unknown (the live read failed, or it succeeded but reported no currently usable window — note says which) or unavailable (this codex/account exposes no quota data) as uncertainty — neither permission nor denial. The account reports only the windows that currently bind it, so primary (shorter/rolling) or secondary (longer) may be null. Read note for plain-language caveats before relying on it.
  3. Do not make a paid call when rate_limit.status is blocked. This is the exception to advisory: the backend reports a spend control (spend_control_reached: true) and the call will fail, so deferring does not help — no quota reset clears it. Stop and tell the user spending is administratively blocked on their Codex account.
  4. Select one route below and load only its needed reference. Use a free dry-run when one exists, and read what it does not cover (see Data exposure).
  5. Declare the paid-call cap before the first active call, then stay within it.
  6. Branch on ok, then on the concrete tool/result type. Verify claims before acting.

Route the request

Situation Tool or workflow Read
One answer, design critique, or second opinion codex_consult active workflows
Stuck mid-debugging or choosing between viable approaches codex_consult active workflows
Review changes already represented in git codex_review_changes active workflows
Proposed implementation diff from an isolated worktree codex_delegate active workflows
A consult, review, or delegate that can exceed the synchronous deadline — high-reasoning-effort or broad repo-grounded work, a multi-file or whole-branch review, or a substantial implementation task matching _async tool background jobs
Move the Claude session into a resumable Codex thread codex_transfer session transfer
Claude and Codex attempt independently, then synthesize independent two-member attempt independent attempt
Claude drafts, Codex critiques, Claude revises declared review–revise review–revise
Optional parameters, idempotency, or a tool error the tool already selected options and errors
MCP server unavailable limited read-only CLI fallback server-down fallback
None of these, or a Codex call would not change the decision no call — proceed without Codex

Read the full file on GitHub · 228 lines

Files

What ships with it

8 files 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.

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. 9d ago First seen · 228 lines · 141 tokens per session scan A e37133130950

Subscribe to this mod's changes

collaborating-with-codex is a skill published in the GitHub repository briandconnelly/codex-in-claude (3 stars, last pushed 7d ago), licensed MIT. It adds 141 tokens to every session and 3,989 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). 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

writing-release-notes

Use when preparing, refreshing, backfilling, or redrafting a release-notes page under docs/releases/ before opening a normal review pull request.

pvliesdonk/markdown-vault-mcp · 35 tokens

skills

Generate a complete SVG file from a natural language description using SEP-1577 multi-step sampling.

sandraschi/inkscape-mcp · 0 tokens

researching-references

Use when a change depends on how something outside this repository actually behaves — a markdown dialect, a file format, git, a protocol, a vendor API — and docs/design/reference/ has no current reference for it, or the one it has is past its review date. Researches from primary sources and writes a dated, versioned…

pvliesdonk/markdown-vault-mcp · 99 tokens

repository-protection

Use when changing branch or tag rulesets, required checks, or the bootstrap workflow: how repository protection is applied and kept in sync.

pvliesdonk/markdown-vault-mcp · 31 tokens

vault-workflow

Use when the user asks you to search, read, or reason about notes in their markdown vault — guides when to use which vault tool and how to chain them for good results.

pvliesdonk/markdown-vault-mcp · 40 tokens

pr-alignment-loop

Iterate a PR to its final form by running two opposing reviewer droids (reviewer-robustness and reviewer-minimalist) in a bounded back-and-forth loop. The main orchestrator synthesizes their feedback, makes edits, runs tests, and stops when both approve, on stagnation, or after 3 rounds. Use when the user asks to…

nikships/skills-registry · 122 tokens