Skip to content
Get packs

Codex-First Delegation Routing

Routes hands-on coding work from Claude Code to Codex workers with fixed model, speed, and fallback rules.

Source Download skill
Repository
steipete/agent-scripts
Stars
7,433
Licence
MIT
Files
6 files

What it does

This skill is a detailed playbook for Claude Code that tells it how to hand hands-on coding work to Codex (a separate coding agent) and keep planning and verification with Claude. It sets exact launch commands and model settings for Codex workers, and specifies fallback rules when rate limits or capacity errors occur, plus guidance for resuming, stopping, and safely restarting workers. Security audits are routed to a different model, and code review is preferably run through a Codex-backed reviewer.

Great for

  • Developers who already run Claude Code alongside Codex and want a consistent delegation workflow
  • Operators managing many parallel Codex workers who need rate-limit fallback and recovery handling
  • Maintainers who want a documented, repeatable review routine via Codex-backed autoreview

Skip it if

  • People without Codex or OpenAI API access, since most instructions assume those tools are installed and configured
  • Anyone expecting a general-purpose coding skill; this is specific to one operator's setup, model names, and wrapper scripts, which may not exist elsewhere
  • Users who run the commands without reading them: the launch flags bypass approvals and sandboxing (--yolo / --dangerously-bypass-approvals-and-sandbox), which grants workers broad execution rights

Try asking

  • “Delegate implementing this feature to a Codex worker on a separate git worktree using the standard launch flags”
  • “Resume a stalled Codex worker session after a rate-limit error, falling back to the alternate model”
  • “Stop a running Codex worker safely and redirect it with a new order file”
  • “Run a security audit lane on the designated audit model”
Step-by-step guide: how to use Codex-First Delegation RoutingInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 6 original files, the MIT licence and an attribution note.

    codex-first.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip codex-first.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip codex-first.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Read the launch flags and the wrapper script before running anything; the flags bypass approval and sandbox checks, so only use them in environments you control.
  • Verify the served service tier with a direct probe rather than trusting rollout logs, as the file notes they record only the requested tier.
  • Do not edit running wrapper scripts in place; follow the write-then-rename procedure described in the file.

What's inside

  • references/coordination.md
  • references/launch-details.md
  • references/resume-and-watchdog.md
  • references/troubleshooting.md
  • scripts/codex-resume.test.sh
  • SKILL.md

Redistributed unmodified from steipete/agent-scripts under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.