expo/skills

Expo Skills – AI‑agent extensions for the Expo ecosystem

What it is

• A collection of skills (metadata files and helper scripts) that teach generic AI coding assistants (Claude Code, OpenAI Codex, Cursor, GitHub Copilot, Gemini, etc.) how to work with Expo, the React‑Native framework for building cross‑platform mobile and web apps.
• The skills encode the “source‑of‑truth” knowledge from the official Expo docs, CLI, and EAS (Expo Application Services) so the agent can answer concrete questions, generate correct configuration files, and even trigger real builds.

How you get it

• Plugin route – for Claude Code or OpenAI Codex you install an official plugin from their marketplaces (claude plugin install expo@claude-plugins-official or codex plugin add expo@openai-curated). The plugin bundles the skills and keeps them up‑to‑date automatically.
• Skills‑CLI route – for any other AI‑coding tool you run the generic skills CLI: npx skills@latest add expo/skills --skill '*'. The CLI copies the SKILL.md files into the agent’s plugin directory and can target a specific agent with --agent <name>.
• Updates are handled via the marketplace for plugins, or with npx skills@latest update for the CLI.

What the skills cover

Category	Example skills	What they help the agent do
Start Here	expo-overview	Provide a high‑level answer when the user’s request is vague – e.g., “What does Expo need to build a native app?”
Framework (free, open‑source)	expo-router, expo-tailwind-setup, expo-animation, expo-native-ui, expo-data-fetching, …	Generate correct folder structures, navigation code, Tailwind/NativeWind config, animation setups, data‑fetching patterns, etc.
Services (paid EAS features)	eas-app-stores, eas-hosting, eas-workflows, eas-update-insights	Write eas.json profiles, create CI/YAML for EAS, query build status, configure hosting, inspect update health metrics.
Experimental	expo-migrate-module	Guide the agent through migrating an Expo native module to the newer macro‑API.

How an agent uses a skill

1. You ask the agent a concrete Expo‑related request (e.g., “Create an EAS workflow that builds previews on pull requests.”).
2. The agent scans the installed SKILL.md files, picks the one whose description matches the task (eas-workflows in this case), and follows the step‑by‑step instructions encoded there.
3. If the skill needs live data (e.g., current docs, triggering a build), the Expo MCP server can be consulted – a lightweight backend that lets the agent run expo install, start EAS builds, fetch crash logs, or take a screenshot of a remote simulator.

Telemetry & feedback

• Telemetry is off by default; when enabled it only sends the skill name, platform, and a hashed install ID – no code or personal data.
• Users can toggle it with EXPO_SKILLS_TELEMETRY=1/0 or by asking the agent to “enable/disable Expo skills telemetry”.
• Each skill includes a submit-expo-feedback command for manual, actionable feedback.

Why it matters

• Building Expo apps involves many moving parts (CLI flags, EAS credentials, platform‑specific quirks). By codifying that expertise into reusable AI‑agent skills, developers can get accurate, up‑to‑date guidance from any supported coding assistant, reducing trial‑and‑error and speeding up onboarding.

License: MIT (free to use and modify).