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

Codex Exec Delegation

Delegate tasks to the OpenAI codex CLI non-interactively from Claude Code, with hardened rules for timeouts and session recovery.

Dev & CodingAdvanced★ 404⑂ 32AI score 9/10Last updated: Sep 27, 2026

What it does

  • Runs codex exec to hand a whole task to the codex CLI — progress on stderr, final result on stdout.
  • Enforces < /dev/null on non-piped calls so codex doesn't hang forever on a never-EOF stdin pipe inside subagent/subprocess contexts.
  • Routes large or untrusted context (diffs, file contents, specs, tool output) through a written file piped over stdin so backticks, $, and quotes are never shell-interpreted.
  • Defines recovery paths for every failure mode: 600s force-backgrounding, hard kill (exit 143), and runs that exit 0 while asking for authorization — capture the session id, resume, or recover the ID from ~/.codex/sessions/.../rollout-*.jsonl.
  • Adds guidance on --sandbox mode selection, absolute randomized -o output paths, XML-tagged prompt shaping, and parallel spawn_agent/wait_agent fan-out.

Who it's for

  • Developers running both Claude Code and the OpenAI codex CLI who want to offload heavy analysis or implementation.
  • Anyone building multi-agent pipelines with parallel sub-agents.
  • Anyone repeatedly bitten by codex exec hanging or returning empty output.

Example uses

  1. "Have codex analyze this repo's auth flow" → runs with --sandbox read-only and returns a grounded report.
  2. Reviewing a long diff: write the diff to a file and pipe it (cat context.txt | codex exec ...) so nothing is interpreted on the way in.
  3. A run that blows past the 10-minute ceiling gets force-backgrounded — re-read the output file once the task notification fires, or resume <session-id> if it was hard-killed.

· · · 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/tobihagemann/turbo/HEAD/claude/skills/codex-exec/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 claude/skills/codex-exec folder from the GitHub repo tobihagemann/turbo into my ~/.claude/skills/codex-exec/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/tobihagemann/turbo.git && mkdir -p ~/.claude/skills && cp -r turbo/claude/skills/codex-exec ~/.claude/skills/

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

  1. Make sure the OpenAI codex CLI is installed and authenticated: run codex --version.
  2. Clone the repo: git clone https://github.com/tobihagemann/turbo.git
  3. Create the skills directory: mkdir -p ~/.claude/skills
  4. Copy the skill in: cp -r turbo/claude/skills/codex-exec ~/.claude/skills/
  5. Verify that references/parallel-execution.md came along with it.
  6. Restart Claude Code and try a trigger phrase like "run codex exec to analyze this file".
  7. Note: this skill requires disabling the Bash sandbox for every codex call — use it only in project directories you trust.
View source on GitHub ↗License: MIT