MCP Server
Give compatible AI assistants access to PrimeVue component documentation, guides, examples, and API metadata.
Introduction #
Model Context Protocol (MCP) lets AI assistants call external tools. The @primevue/mcp package uses those tools to search and read the PrimeVue documentation included with the package.
- Find components, guides, examples, and setup instructions.
- Read component APIs and documentation sections.
- Check component usage against documented properties and events.
- Return links to the related PrimeVue documentation when available.
The server requires Node.js 22 or newer. It is read-only and does not change your project files or PrimeVue configuration.
CLI Setup #
The recommended setup is the PrimeVue Plugin. It installs this MCP server together with the PrimeVue skills that guide common implementation, setup, theming, migration, and troubleshooting tasks.
Replace copilot with claude, codex, gemini, or cursor for another supported assistant. Use Manual Setup when you want the MCP server without the plugin skills.
Manual Setup #
Claude Code #
Add the MCP server to your user configuration to use it in every project, or omit -s user to add it only to the current project. Start a new Claude Code session after registration.
See Claude Code MCP Documentation for more details.
VS Code #
Create .vscode/mcp.json for project-level setup. To use the server across projects, add the same servers entry to your VS Code user MCP configuration.
See VS Code MCP Documentation for more details.
OpenAI Codex #
Add the MCP server with the Codex CLI, or add the same configuration to ~/.codex/config.toml.
See OpenAI Codex MCP Documentation for more details.
Cursor #
Create .cursor/mcp.json in your project or ~/.cursor/mcp.json for global configuration.
See Cursor MCP Documentation for more details.
Tools #
The PrimeVue MCP server provides eight tools.
| Tool | Purpose |
|---|---|
| list | List available components, guides, examples, resources, or modes. |
| search | Search generated component, guide, section, API, and example text. |
| get_component | Read component documentation, API metadata, and available sections. |
| get_guide | Read a generated PrimeVue guide and its source metadata. |
| get_example | Return an example from a documented component section. |
| get_setup | Return setup guidance for a supported PrimeVue environment. |
| validate_usage | Validate component usage against generated PrimeVue API metadata. |
| version | Report the server version and the status of its documentation data, tools, and resources. |
Data and Validation #
The MCP package includes a generated copy of the PrimeVue documentation. This keeps the server usable without a PrimeVue source checkout and makes its answers match the installed package version. Results include a primeui://primevue resource URI and, when available, a link to the public documentation.
- The tools do not edit files, install dependencies, or update configuration.
- validate_usage checks documented component metadata. If the required metadata is unavailable, it reports that limitation instead of guessing.
- Run your project type checks, tests, and browser checks after making changes. MCP validation does not replace them.
If you installed the PrimeVue Plugin, do not register the MCP server again with a manual configuration. Run the CLI doctor command if the server is missing or appears more than once.