Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
Evgeny Gurin avatar

Autopilot Codex Marketplace

  • Updated May 6, 2026
  • evgenygurin/autopilot-codex

autopilot-codex-marketplace is a Claude Code plugin marketplace where Codex acts as engineering manager so developers can run a plan-review-implement loop with Claude Code as implementer.

About

autopilot-codex-marketplace distributes a Claude Code plugin where OpenAI Codex plays senior engineering manager and Claude Code is the implementer. Developers invoke /autopilot-codex:autopilot with a goal to get a persisted plan, per-edit cross-model reviews, and a Stop-hook loop that continues until Codex marks the work done. It suits teams wanting GPT-Claude cross-checking without API key cost when using a ChatGPT subscription.

  • Codex decomposes goals into ordered tasks with acceptance criteria; Claude implements
  • PostToolUse hook sends every Edit/Write to Codex for approve, fix, or reject verdicts
  • Stop hook asks Codex done? and hands next task until complete or iteration cap
  • Bash safety rails block destructive commands; works with ChatGPT subscription via codex login

Autopilot Codex Marketplace by the numbers

  • Data as of Jul 7, 2026 (Skillselion catalog sync)
/plugin marketplace add evgenygurin/autopilot-codex

Add your badge

Show developers this marketplace is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Last updatedMay 6, 2026
Repositoryevgenygurin/autopilot-codex

How do I automate a manager-implementer loop where another model plans tasks and reviews every diff while Claude Code writes the code?

Let OpenAI Codex act as engineering manager—plan tasks, review diffs, and drive a Stop-hook loop while Claude Code implements.

Who is it for?

Developers with Codex CLI and a ChatGPT subscription who want cross-model review and autonomous task loops inside Claude Code.

Skip if: Teams without Codex CLI access or users who prefer single-model workflows without manager-implementer separation.

What you get

A markdown plan and state.json track tasks; each edit gets a Codex verdict; the loop advances until done, abort, or the iteration cap.

Recommended Marketplaces

FAQ

Do I need an OpenAI API key?

No—a ChatGPT subscription with codex login is the recommended path; API key login is also supported.

What triggers Codex review?

Every Edit or Write triggers a PostToolUse hook that asks Codex to approve, request fixes, or reject against the current task acceptance criteria.

How does the loop stop?

The Stop hook asks Codex for a done verdict; you can also use /autopilot-codex:codex-stop or hit the default 30-iteration cap.

AI Agentsagentsautomation

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.