# Packary

> Trusted AI agent skill platform.

Skills your agent can trust.

## Connect an agent

- Remote HTTP MCP: `https://packary.xyz/mcp`
- MCP server name: `packary`
- One-command local setup: `curl -fsSL https://packary.xyz/install.sh | sh`
- Standard Agent Skills index: [https://packary.xyz/.well-known/skills/index.json](https://packary.xyz/.well-known/skills/index.json)

Once connected, call `search_skills`, inspect a match with `get_skill`, and execute it with `run_skill`. If the client cannot add remote HTTP MCP servers, install Packary with the command above and use `packary search`, `packary info`, and `packary run`.

## Discovery

- [Registry API](https://packary.xyz/api/v1/skills)
- [OpenAPI](https://packary.xyz/openapi.json)

## Skills

- [hello-ai-world](https://packary.xyz/skills/hello-ai-world.md): Print a friendly greeting from an installed AI skill
- [use-packary](https://packary.xyz/skills/use-packary.md): Help an AI agent discover and run trusted Packary skills
- [review-code-changes](https://packary.xyz/skills/review-code-changes.md): Catch real defects before code ships
- [write-release-notes](https://packary.xyz/skills/write-release-notes.md): Turn commits into release notes people can scan
- [plan-product-launch](https://packary.xyz/skills/plan-product-launch.md): Build a focused launch plan with owners and metrics
- [count-project-files](https://packary.xyz/skills/count-project-files.md): Count regular files in the current project safely
- [analyze-product-copy](https://packary.xyz/skills/analyze-product-copy.md): Analyze product copy for clarity, specificity, and a strong next action. Use when an agent needs to critique a headline, product description, landing-page paragraph, or call to action.