awesome-codex-skills is a curated index of skills for OpenAI’s Codex, described on the repository as “A curated list of practical Codex skills for automating workflows across the Codex CLI and API.”
Unlike most awesome-lists, this one is partly a package: roughly 48 of its entries are folders in the repository itself, with about 12 pointing to external repositories. It ships a Python installer, so it functions as a small distribution channel rather than only a set of links.
How Codex skills actually load
The README explains the mechanic that makes the format work, and it is worth knowing because it shapes how a skill should be written: each skill is a folder with a SKILL.md holding metadata and step-by-step guidance, and Codex “reads the metadata to decide when to trigger a skill and loads the body only after it fires.”
That two-stage loading is the whole economy of the format. The metadata is always in context; the instructions are not. A skill with a vague description never fires, and a skill with a bloated body costs nothing until it does. If you write your own, that sentence is the design brief.
What is in it
About 61 entries, with the categorised 60 breaking down as:
- Productivity & Collaboration — 17
- Development & Code Tools — 16
- Meta & Utilities — 10
- Data & Analysis — 9
- Communication & Writing — 8
The spread is notable: this is not a developer-only list. More than half sits in productivity, communication and data work, which matches how Codex is actually used outside engineering teams.
Running the installer
Clone, then run the installer against the skill path you want:
git clone https://github.com/ComposioHQ/awesome-codex-skills.git
cd awesome-codex-skills
python skill-installer/scripts/install-skill-from-github.py \
--repo ComposioHQ/awesome-codex-skills --path meeting-notes-and-actionsThe installer “places it in $CODEX_HOME/skills/<skill-name>” and you then “Restart Codex to pick up new skills.” Manual installation is just copying the folder into $CODEX_HOME/skills/ — default ~/.codex/skills/ — and restarting.
The same installer handles external skills from other repositories:
python3 ~/.codex/skills/.system/skill-installer/scripts/install-skill-from-github.py \
--repo hyhmrright/brooks-lint --path skills/brooks-lint --name brooks-lintTo confirm what landed, the README suggests ls ~/.codex/skills and head ~/.codex/skills/<skill>/SKILL.md. Reading that head output is a good habit regardless — it is the metadata Codex will act on.
Two gaps to know about
No licence is stated. The repository has no License section. For an index that is awkward; for a repository that contains 48 skill folders you are copying onto your machine and into your workflow, it matters more. Without a stated licence, the default position is that you have no explicit permission to redistribute or relicense that content. For personal use it is unlikely to bite; for anything you ship to a team or a client, ask before building on it.
Review is undocumented. The contribution guidance is “PRs welcome” with a request to “Add real, reusable skills, keep descriptions precise.” There is no described review process or quality bar. Since a skill is instructions your agent will follow, read the SKILL.md before installing — the same discipline you would apply to a shell script someone sent you.
External entries compound both points: they live in other repositories, with their own terms and their own maintenance, and this index does not cover them.
Worth it for
Finding Codex skills that are not developer-centric, and for the installer, which removes the fiddly part of getting a folder into the right place. The category split is the real signal here — if you want Codex doing meeting notes or data chores rather than refactors, this is where that work has been collected.
Install one skill, read its SKILL.md end to end, and see whether the trigger metadata fires when you expect. That tells you more about the quality of a list like this than its entry count does.
