TanStack MCP Server
[](https://lobehub.com/mcp/zpeppoz-tanstack-mcp)
# @g7aro/tanstack-mcp
MCP server that wraps the [TanStack CLI](https://tanstack.com/cli) to provide programmatic access to TanStack documentation, libraries, add-ons, ecosystem partners, and project scaffolding.
Built as a drop-in replacement after TanStack [removed the built-in MCP server](https://tanstack.com/cli/latest/docs/mcp-migration.md) from `@tanstack/cli`.
## Tools
| Tool | Description |
|---|---|
| `listTanStackAddOns` | List all available TanStack Start add-ons for a given framework |
| `getAddOnDetails` | Get detailed info about a specific add-on (files, deps, options, routes) |
| `createTanStackApplication` | Scaffold a new TanStack Start app with add-ons and options |
| `tanstack_list_libraries` | List all TanStack libraries with descriptions and links |
| `tanstack_doc` | Fetch the full content of a TanStack documentation page |
| `tanstack_search_docs` | Search across TanStack documentation |
| `tanstack_ecosystem` | List TanStack ecosystem partners with optional filters |
## Quick Install
Auto-detect installed AI clients and register the MCP server in one command:
```bash
npx @g7aro/tanstack-mcp --install
```
This will detect and configure all supported clients on your machine.
### Options
```bash
npx @g7aro/tanstack-mcp --install # Interactive — pick which clients
npx @g7aro/tanstack-mcp --install --all # Install into all detected clients
npx @g7aro/tanstack-mcp --install cursor codex # Install into specific clients only
npx @g7aro/tanstack-mcp --uninstall # Remove from all clients
```
### Supported clients
| Client | Detection | Config method |
|---|---|---|
| Claude Code | `claude` CLI | `claude mcp add` |
| Codex (OpenAI) | `codex` CLI | `codex mcp add` |
| Cursor | `~/.cursor/` dir | `~/.cursor/mcp.json` |
| Windsurf | `~/.windsurf/` dir | `~/.windsurf/mcp.json` |
| Trae | `~/.trae/` dir | `~/.trae/mcp.json` |
| Antigravity | `~/.gemini/antigravity/` dir | `mcp_config.json` |
| OpenCode | `~/.config/opencode/` dir | `opencode.json` |
| Zed | `~/.config/zed/settings.json` | `settings.json` |
| VS Code (Copilot) | `settings.json` / `code` CLI | `settings.json` |
## Manual Setup
If you prefer to configure manually, add to your client's MCP config:
```json
{
"mcpServers": {
"tanstack": {
"command": "npx",
"args": ["-y", "@g7aro/tanstack-mcp"]
}
}
}
```
## Tools
| Tool | Description |
|---|---|
| `listTanStackAddOns` | List all available TanStack Start add-ons for a given framework |
| `getAddOnDetails` | Get detailed info about a specific add-on (files, deps, options, routes) |
| `createTanStackApplication` | Scaffold a new TanStack Start app with add-ons and options |
| `tanstack_list_libraries` | List all TanStack libraries with descriptions and links |
| `tanstack_doc` | Fetch the full content of a TanStack documentation page |
| `tanstack_search_docs` | Search across TanStack documentation |
| `tanstack_ecosystem` | List TanStack ecosystem partners with optional filters |
## Prerequisites
- Node.js >= 18
- `npx` available in PATH (ships with npm)
## How it works
Each MCP tool maps to a `@tanstack/cli` command with `--json` output:
```
listTanStackAddOns -> tanstack create --list-add-ons --framework <f> --json
getAddOnDetails -> tanstack create --addon-details <id> --framework <f> --json
createTanStackApplication -> tanstack create <name> --framework <f> --add-ons <a,b> ...
tanstack_list_libraries -> tanstack libraries --json
tanstack_doc -> tanstack doc <library> <path> --json
tanstack_search_docs -> tanstack search-docs "<query>" --json
tanstack_ecosystem -> tanstack ecosystem --json
```
The server spawns `npx @tanstack/cli` for each invocation, parses the JSON output, and returns it through the MCP protocol over stdio.
## Development
```bash
npm install
npm run build # compile TypeScript -> dist/
npm start # run the server (stdio)
npm run dev # watch mode
```
## License
MIT
TDQS
Scored across 7 tools
Each tool has a distinct purpose: scaffolding (create, list add-ons, get add-on details), documentation (fetch page, search, list libraries, list ecosystem partners). No overlap.
First three tools use camelCase (createTanStackApplication, getAddOnDetails, listTanStackAddOns) while last four use lowercase underscores (tanstack_doc, etc.). Also, the first three include verb-noun pattern but the last four lack clear action verbs in names (e.g., tanstack_doc is just a noun).
Seven tools is well-scoped for a server covering TanStack Start scaffolding and documentation reference. Not too many, not too few.
The set covers key scaffolding operations (create, list, get details) and comprehensive documentation access (search, fetch, list libraries, list ecosystem). No obvious gaps for the intended domain.