by NulightJens
Run Claude Fable 5 + OpenAI Codex as co-founders in one Claude Code skill. Fable is the Visionary (plan, standards, review), Codex is the Integrator (filter, build, report). The Rocket Fuel V/I operating system: interview, adversarial Same Page Meeting, 3-7 Rocks, sandboxed build, Level 10 review.
# Add to your Claude Code skills
git clone https://github.com/NulightJens/rocket-fuel-skillGuides for using ai agents skills like rocket-fuel-skill.
rocket-fuel-skill is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by NulightJens. Run Claude Fable 5 + OpenAI Codex as co-founders in one Claude Code skill. Fable is the Visionary (plan, standards, review), Codex is the Integrator (filter, build, report). The Rocket Fuel V/I operating system: interview, adversarial Same Page Meeting, 3-7 Rocks, sandboxed build, Level 10 review. It has 53 GitHub stars.
rocket-fuel-skill's catalog security scan is still queued. You can run an instant dependency and prompt-injection check now with the "Scan for vulnerabilities" button above.
Clone the repository with "git clone https://github.com/NulightJens/rocket-fuel-skill" and add it to your Claude Code skills directory (see the Installation section above).
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh rocket-fuel-skill against similar tools.
No comments yet. Be the first to share your thoughts!
Unlocks once the catalog security scan passes (runs nightly).
The deep catalog scan for this skill is still queued. Run an instant dependency check now instead.

One sees the future. One makes it happen. Run Claude Fable 5 and OpenAI Codex as co-founders in a single Claude Code command.
Fable 5 takes the Visionary seat: it interviews you, writes the plan, sets the standards, and reviews the result. Codex takes the Integrator seat: it filters the plan, kills scope creep, builds the code in its own sandbox, and reports honestly. They argue in a bounded meeting until the verdict line says SAME PAGE, and only then does anyone write code. Based on the Visionary/Integrator operating system from Rocket Fuel by Gino Wickman and Mark C. Winters.
You make exactly three decisions per run: the interview answers, rare deadlock tie-breaks, and the final commit. The skill guides you through everything else: it tells you which mode it detected, what happens next, and where every artifact lands.
npm i -g @openai/codex@latest && codex login # ChatGPT account works, no API key
git clone https://github.com/NulightJens/rocket-fuel-skill.git ~/rocket-fuel-skill
mkdir -p ~/.claude/skills
ln -s ~/rocket-fuel-skill/rocket-fuel ~/.claude/skills/rocket-fuel
Restart Claude Code once. Requirements: Claude Code, Codex CLI 0.130+, git.
One command, plain English. The skill picks the right mode from context:
| You say | You get |
|---|---|
/rocket-fuel build me a habit tracker |
Full kickoff: interview → adversarial plan review → Codex builds → Fable verifies |
/rocket-fuel this repo is a mess, clean it up |
Codebase audit → ranked issue list → you pick 3-7 fixes → build |
/rocket-fuel review PLAN.md |
Codex attacks your plan read-only until it approves |
/rocket-fuel have codex add dark mode |
One scoped task, built by Codex, diff-reviewed and proof-run by Fable |
From there the skill walks you through it: one interview question at a time with a recommended answer attached, live verdicts from Codex each round, and a clear prompt before anything is committed. The full argument between the two models is saved to SAME-PAGE-LOG.md in your repo, next to PLAN.md.
Based on the Visionary/Integrator operating system from Rocket Fuel by Gino Wickman and Mark C. Winters. Built by @NulightJens.
MIT License.