Build Loop Codex

A structured workflow for building features with OpenAI's Codex CLI that enforces a build, review, test, fix cycle before marking anything done. It works from either a plan file with checkboxes or a direct prompt, runs Codex's /review command on every change to catch bugs and style issues, then makes you verify the feature end to end before moving on. The discipline is the point: no skipping review, no untested code, no silent scope creep. If you're the kind of person who ships half-tested work when you're in a hurry, this will slow you down on purpose. If you want every commit to actually work, it's a forcing function that keeps you honest.

skills.sh installs: 1.5K
GitHub stars: 233
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add buildgreatproducts/builder-os --skill build-loop-codex --agent claude-code

Installs into .claude/skills of the current project.

Files

Codex Build Loop

Quality-gated feature work: nothing ships on "it compiles" — every increment is built, reviewed, tested end to end, and fixed before the user hears "done."

Source of work

A plan file exists (roadmap, refactor plan, or task list with - [ ] checkboxes — search the repo): work the first unchecked task. Tasks are ordered intentionally — never skip ahead. If the plan references spec docs, read only the sections relevant to the current task.
No plan (or the request is outside it): build from the user's prompt. Restate it as a verifiable goal with 2–4 success criteria and confirm scope in one message before building.

The loop

Run per task (or per prompted feature). Do not advance until every step passes.

Build. Implement exactly what the task specifies. Simplest implementation that satisfies it, surgical changes, no speculative scope. Match existing project conventions.

SKILL.md on GitHub