CodexSkills skill
A Codex skill collection: phased planning, parallel subagents, documentation access, a browser, 51 hooks and ready agent roles.
- 1,041 ★ on the repository
- Last commit
Extended access
Why this level
- Subagents of the `parallel-task` skill edit project files simultaneously, and hooks run scripts on Codex events.
- Some skills send code and queries to third-party providers and require their keys.
Install
- Works in
Manual install
npx skills add am-will/codex-skills --listThe first README command: lists skills without installing. Install with `npx skills add am-will/codex-skills --skill planner -g`.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Subagents of the `parallel-task` skill edit project files simultaneously, and hooks run scripts on Codex events.
- Some skills send code and queries to third-party providers and require their keys.
This is third-party code. Review the repository files before installing.
What it does
CodexSkills gathers the author's Codex skills in one repository. The `planner` skill builds a phased plan of sprints and atomic tasks, `plan-harder` digs deeper into a task, and `parallel-task` runs several subagents at once on a ready plan. `llm-council` asks Claude, Codex and Gemini for independent plans, after which a judge picks the best approach. Documentation comes through `ctx7old` via Context7, `openai-docs-skill` via the OpenAI documentation MCP server and `read-github` via gitmcp.io. For the browser there are `agent-browser` and `gemini-computer-use`. The `create-hook`, `pluginstaller` and `role-creator` skills create hooks, install Codex plugins and describe agent roles in TOML. Separate directories hold 51 ready hooks and a set of multi-agent configurations.
Who it is for. Codex CLI users who want a planner, parallel executors and a set of hooks in one bundle.
Good fit when
- A big feature must be split into sprints with atomic tasks and handed to subagents.
- One plan is not enough and you want to compare the solutions of three models through a judge.
- The agent must check current documentation of a library or repository.
- You need a hook that blocks dangerous commands or checks commits and messages.
- The project needs a separate agent role described in TOML.
Not a fit when
- You need one narrow capability: the full set is excessive, install a single skill instead.
- You have no Context7 or Gemini keys, or subscriptions to the models `llm-council` needs.
- Requests and code must not go to third-party providers.
Example request
Use planner to draft an OAuth plan by sprints, then run parallel-task and execute the independent tasks in parallel.Limitations
The `ctx7old` skill requires a key in `CONTEXT7_API_KEY`, `gemini-computer-use` requires `GEMINI_API_KEY`, and `llm-council` needs keys or subscriptions for Claude, OpenAI and Google plus a run of `./setup.sh`. The `parallel-task` skill works only with a ready plan from `planner`. The repository has no license file. Hooks are adapted to the current Codex event model and may go stale along with it. Some skills, such as `frontend-design`, are imported from Anthropic and Vercel. The last commit was made in July 2026.
How to disable. Run `npx skills remove <name>` for each installed skill. Remove hooks from the project's `.codex/hooks.json` and `.codex/hooks/`, and delete agent roles from your TOML files.
FAQ
How do I install only the planner?
Run `npx skills add am-will/codex-skills --skill planner -g`. The `-g` flag installs the skill for all projects, without it the skill goes into the current one. Pick agents with `-a`, for example `-a claude-code -a codex`.
How does planner differ from plan-harder?
`planner` builds a phased plan with sprints and atomic tasks. `plan-harder` is described as an enhanced variant with more detailed analysis and breakdown.
What does parallel-task need?
A ready plan file created by `planner`. The skill launches several subagents and each takes its own task from the plan.
How do I install a hook from the catalog?
Run `python3 hooks/aitmpl-codex/install-bundle.py hooks/aitmpl-codex/<category>/<bundle> <project-folder>` or copy the bundle's `hooks.json` into the project's `.codex/hooks.json` by hand and restart Codex.
Which browser skill should I choose?
One is enough. The author recommends `agent-browser` for speed and simplicity, while `gemini-computer-use` drives a browser through Gemini and Playwright with a safety confirmation loop.
Discussion
No questions or reviews yet.
Related
SkillsA Codex plugin of five independent meta-skills and one tldraw skill: improves prompts, goals, research, plans and commits.
A base AGENTS.md and 12 skills: delegation to Codex, Grok, Devin and Antigravity, planning, copywriting, scoped changes.
Skill setting up Markdown file memory for Codex via AGENTS.md: profile, rules, lessons, errors and capability requests.
A catalog of nearly two hundred skills: code delivery, review, debugging, AI governance, product work and agent coordination.
Agents, skills and templates for structured AI development in OpenCode, Codex and Claude Code: docs, plans, review, checkpoints
A skill that links two to four open Codex CLI windows: one splits the work while the others execute, cross-check and deliver.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| CodexSkills (this card) | no | 1,041 | |
| Awesome Codex Skills | no | 17,302 | a Composio community list of skills for bug triage and CI fixes; here one author collects a planner, subagents, hooks and roles |
| Claude-Codex Skills | no | 594 | skills for the whole development cycle from product to operations; CodexSkills focuses on planning, parallel agents and Codex hooks |
README in short
The README groups skills by area: agent orchestration, documentation access, prompting, frontend, Codex tooling and browser. After the list come installation through the skills CLI, a catalogue of 51 hooks by category and a catalogue of ready multi-agent roles.