CodexSkills

A collection of Codex/agent skills for planning, documentation access, frontend development, and browser automation.

Available skills

Agent Orchestration

• planner: Create comprehensive, phased implementation plans with sprints and atomic tasks. Use for planning implementations, breaking down features, or creating structured roadmaps.
• plan-harder: Enhanced planning variant for more detailed analysis and task breakdown.
• parallel-task: Execute plan files by launching multiple parallel subagents to complete tasks simultaneously. Requires an existing plan file from planner.
• llm-council: Multi-agent orchestration system for planning complex tasks. Spawns multiple AI planners (Claude, Codex, Gemini) to generate independent plans, then uses a judge agent to synthesize the best approach. Includes a real-time web UI for monitoring progress and refining plans interactively.

Documentation Access

• ctx7old: Fetch up-to-date library documentation via Context7 CLI.
• openai-docs-skill: Query OpenAI developer docs via the OpenAI Docs MCP server using CLI.
• markdown-url: Prefix any website you need to visit with https://markdown.new/ for a clean, Markdown-friendly view.
• read-github: Read and search GitHub repository documentation via gitmcp.io MCP service. Converts github.com/owner/repo URLs to gitmcp.io/owner/repo for LLM-friendly access to repos.

Frontend & Design

• frontend-design: Distinctive frontend design system guidance (imported from Anthropic).
• frontend-responsive-ui: Responsive UI standards (imported from Anthropic).
• vercel-react-best-practices: React/Next.js performance guidance (imported from Vercel).

Codex Tooling

• create-hook: Create or update Codex hooks for global or project scope. Guides users through scope, hook type, and behavior, then writes hook config and scripts.
• pluginstaller: Install a Codex plugin from a GitHub repo into a repo or personal marketplace. Handles plugin discovery, validation, and marketplace registration.
• role-creator: Create and update custom Codex agents as standalone TOML files. Collects required inputs and writes agent definitions to global or project scope.

Browser Automation

• gemini-computer-use: Gemini 2.5 Computer Use browser-control agent skill (Playwright + safety confirmation loop).
• agent-browser: Fast Rust-based headless browser automation CLI from Vercel Labs with snapshot/act pattern for AI agents.

Installation

Install skills using the skills.sh CLI:

# List available skills before installing
npx skills add am-will/codex-skills --list

# Install specific skills to user scope (global)
npx skills add am-will/codex-skills --skill planner --skill context7 -g

# Install all skills interactively (prompts for selection)
npx skills add am-will/codex-skills -g

# Install to specific agents
npx skills add am-will/codex-skills --skill planner -a claude-code -a codex -g

# Install to current project (instead of global)
npx skills add am-will/codex-skills --skill planner

# Non-interactive install (skip prompts)
npx skills add am-will/codex-skills --skill planner -g -y

CLI Options:

Flag	Purpose
-g, --global	Install to user directory (global for all projects)
-a, --agent <agents...>	Target specific agents (claude-code, codex, etc.)
-s, --skill <skills...>	Install specific skills by name
-l, --list	List available skills without installing
-y, --yes	Skip confirmation prompts

Other commands:

npx skills list          # Show installed skills
npx skills remove <name> # Uninstall a skill
npx skills update        # Update all installed skills

Note on Browser Tools: The repo includes two browser automation tools (gemini-computer-use and agent-browser). You don't need to install both - choose the one that best fits your workflow. I recommend agent-browser for speed and simplicity.

Note on Context7: The ctx7old skill requires a Context7 API key in CONTEXT7_API_KEY. See skills/ctx7old/.env.example for setup.

Note on Gemini Computer Use Skill: This skill requires a GEMINI_API_KEY. Ask Codex to help you set it up.

Note on llm-council: This skill requires API keys or Active Subscriptions for multiple providers (Claude/Anthropic, OpenAI for Codex, Google for Gemini). Run ./setup.sh in the skill directory to configure. Includes a real-time web UI that auto-launches during planning sessions.

---

Codex Hook Catalog

The hooks/ directory contains a catalog of 50 ready-to-install Codex hook bundles, generated from davila7/claude-code-templates and adapted for the Codex event model.

Categories:

Category	Bundles	Examples
automation	18	Slack/Discord/Telegram notifications, Vercel auto-deploy, dependency checker, change logger
development-tools	8	Debug window, lint-on-save, smart formatting, file backup, worktree ghostty
git	3	Conventional commits, prevent direct push, validate branch name
git-workflow	2	Auto git-add, smart commit
monitoring	2	Desktop notification on stop, LangSmith tracing
performance	2	Performance budget guard, performance monitor
post-tool	4	Format JS/Python files, git-add changes, run tests after changes
pre-tool	4	Backup before edit, console-log cleaner, notify before bash, update search year
quality-gates	3	Plan gate, scope guard, TDD gate
security	4	Dangerous command blocker, file protection, secret scanner, security scanner
testing	1	Test runner

Install a bundle:

python3 hooks/aitmpl-codex/install-bundle.py hooks/aitmpl-codex/<category>/<bundle> <target-repo>

Manual install: Copy the bundle's hooks.json into <repo>/.codex/hooks.json and its .codex/hooks/ scripts into <repo>/.codex/hooks/, then restart Codex.

See hooks/README.md for full details on compatibility, testing harness, and bundle structure.

---

Custom Multi-Agents for Codex

The agents/ directory contains custom multi-agent configurations for Codex. These TOML-based agent definitions provide specialized roles (architect, frontend, backend, security, etc.) and worker tiers for parallel task execution. Use them with Codex to spin up purpose-built agents tailored to your workflow.