Claude Skill MartBrowse skillsQuick linesLearn by videoTerminal guideWhat is a Skill?
 
← Back to list

Collaborating with Codex

A bridge skill that lets Claude Code delegate prototyping, debugging, code review, and cross-model second opinions to Codex CLI.

Dev & CodingAdvanced★ 132⑂ 17AI score 9/10Last updated: Sep 13, 2026

What it does

  • Wraps codex exec in JSON mode through scripts/codex_bridge.py, returning structured results (success, SESSION_ID, agent_messages, commands_ran).
  • Supports multi-turn delegation with SESSION_ID or --last, so you can go analyze → propose minimal fix → check edge cases.
  • Encodes an explicit safety model: read-only by default, workspace-write only when writes are warranted, plus separate gates for live web search (--search) and shell network (--network).
  • Documents the host-side approval trap: why the bridge Bash call can be silently denied and which permissions.allow / sandbox.excludedCommands patterns fix it.
  • Ships reference docs for prompt templates, XML prompt blocks, anti-patterns, shell quoting, handoff/parallel workflows, and gpt-image-2 image generation.

Who it's for

  • Developers running Claude Code alongside Codex CLI who want a genuine cross-model second opinion.
  • Teams that want review/diagnosis read-only and implementation confined to an isolated /tmp worktree.
  • Anyone running long agent tasks in the background and monitoring streamed stderr progress.

Example uses

  1. Read-only review of src/auth.py asking for a unified diff only, applied after primary-agent review.
  2. git worktree add -b codex/fix /tmp/wt-fix HEAD, then --sandbox workspace-write for a focused fix plus narrow verification.
  3. Generate a PR review report with codex exec review --base origin/main -o /tmp/codex-review.md.
  4. Tune cost/latency via -c 'model_reasoning_effort="medium"' or fan out with --enable multi_agent.

· · · Install guide · · ·

Try it now, no install

Paste this into Claude to use the skill without installing anything.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/appautomaton/agent-designer/HEAD/skills/collaborating-with-codex/SKILL.md

What I want: (describe your task here)

If Claude can't open the link, open it yourself and paste the contents instead.

↓ If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the skills/collaborating-with-codex folder from the GitHub repo appautomaton/agent-designer into my ~/.claude/skills/collaborating-with-codex/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/appautomaton/agent-designer.git /tmp/agent-designer && mkdir -p ~/.claude/skills && cp -r /tmp/agent-designer/skills/collaborating-with-codex ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Prerequisites: install Python 3 and the Codex CLI, then authenticate (or set OPENAI_API_KEY). Verify with codex --version.
  2. Clone the repo: git clone https://github.com/appautomaton/agent-designer.git /tmp/agent-designer
  3. Copy the skill: mkdir -p ~/.claude/skills && cp -r /tmp/agent-designer/skills/collaborating-with-codex ~/.claude/skills/
  4. Smoke test: python3 ~/.claude/skills/collaborating-with-codex/scripts/codex_bridge.py --help
  5. Probe sandbox health: codex sandbox -- true (exit 0 is healthy; exit 182 means the kernel can't enforce Codex's sandbox).
  6. Add "Bash(python3 *collaborating-with-codex*bridge.py*)" to permissions.allow in your Claude Code settings.json; on sandboxed hosts add the same pattern to sandbox.excludedCommands.
  7. Restart Claude Code and trigger it with something like "delegate a review of this file to Codex."
View source on GitHub ↗