# VueUse MCP MCP server that gives AI coding agents instant access to VueUse documentation — search composables, read docs, and get suggestions. ## How It Works When You Code This is an [MCP (Model Context Protocol)](https://modelcontextprotocol.io) server. When you connect it to an AI coding agent like Claude Code, the agent can automatically look up VueUse documentation while helping you write code. For example, if you ask "help me track the mouse position in Vue," the agent will: 1. Call `search_composables` or `suggest_composable` to find relevant composables 2. Call `get_composable_docs` to read the full documentation for `useMouse` 3. Write code using the correct API, options, and types — without you needing to look anything up You don't need to manually call any tools. The agent decides when to use them based on your questions. ## Quick Start ### Claude Code ```bash claude mcp add vueuse-mcp -- npx vueuse-mcp ``` ### Codex CLI Add to your MCP config: ```json { "servers": { "vueuse": { "command": "npx", "args": ["vueuse-mcp"] } } } ``` ### OpenCode Add to your `opencode.json`: ```json { "mcp": { "vueuse": { "command": "npx", "args": ["vueuse-mcp"] } } } ``` ## Available Tools | Tool | Description | |------|-------------| | `search_composables` | Search by name, keyword, or concept. Returns ranked results with descriptions. | | `get_composable_docs` | Get full docs for a composable — signature, types, options, examples, related functions. | | `suggest_composable` | Describe a use case in natural language and get ranked suggestions with trade-offs. | | `list_categories` | List all categories with composable counts. Optionally filter by package. | ## Keeping Data Fresh The composable data is scraped from the VueUse GitHub repo. To update: ```bash npm run scrape && npm run build ``` Set a `GITHUB_TOKEN` environment variable to avoid rate limits during scraping. See `.env.example`. ## Development ```bash git clone https://github.com/kovoor/vueuse-mcp.git cd vueuse-mcp npm install npm run scrape # fetch composable data from GitHub npm run build # compile TypeScript npm start # run the server ``` ## Contributing See [CONTRIBUTING.md](CONTRIBUTING.md) for how to add tools, update the scrape pipeline, and submit PRs.