Back to directory

Awesome Codex Skills

A curated list of practical skills for automating workflows in the Codex CLI and API, each a self-contained instruction bundle Codex loads on demand.

Awesome Codex Skills

Awesome Codex Skills is a curated list of practical skills for automating workflows across the Codex CLI and API. A Codex skill is a modular instruction bundle living in its own folder with a SKILL.md file that declares a name and description plus step-by-step guidance; Codex reads only the metadata to decide when a skill should trigger, and loads the full body only once it fires, which keeps the context sent to the model small.

Skills can be installed with a Python skill-installer script that fetches a skill directly from GitHub into Codex's skills directory, or copied in manually before restarting Codex to pick up the new metadata. Listed skills are grouped under headings including development and code tools, productivity and collaboration, communication and writing, data and analysis, and meta and utilities; examples include a linting skill that grounds AI code review in cited engineering books across several analysis modes, and a multi-agent orchestrator that runs parallel Codex agents in isolated git worktrees.

The project also documents how to write and contribute a new skill.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.