Is Context7 MCP Server official?
Yes. Context7 and its MCP server are built by Upstash; the source is at github.com/upstash/context7 and the npm package is `@upstash/context7-mcp`.
@upstash
Basic information
Context7 Platform -- Up-to-date code documentation for LLMs and AI code editors
Add this server to your MCP-compatible client using the configuration below.
{
"mcpServers": {
"context7-mcp": {
"command": "npx",
"args": [
"ctx7",
"setup"
]
}
}
}Tool list not available yet
This listing does not currently have a tool list to display. This does not mean the server has no tools; check its documentation for supported capabilities.
Fetch the live tool list by running this server in a temporary sandbox using the button above.
Context7 MCP Server is Upstash's official Context7 MCP server: it pulls up-to-date, version-specific library documentation and code examples into your AI coding assistant's context, so answers use current APIs instead of outdated training data.
Context7 is built by Upstash. The MCP server source lives in the upstash/context7 repository (MIT licensed) and is published on npm as @upstash/context7-mcp; the API backend, parsing and crawling engines that build the documentation index are private. You can use it as a hosted remote server or run it locally, and it also has a CLI + Skills mode that works without MCP.
The fastest route is the setup CLI, which needs Node.js 22.13 or newer:
npx ctx7 setup
It signs you in with OAuth, generates an API key, and lets you choose between CLI + Skills mode and MCP mode. Add --cursor, --claude or --opencode to target a specific agent. To undo it later, run npx ctx7 remove.
For a manual setup, use the remote server URL https://mcp.context7.com/mcp and send your key in an Authorization: Bearer YOUR_API_KEY header. A free API key from context7.com/dashboard is recommended for higher rate limits. To run it locally instead, use npx -y @upstash/context7-mcp --api-key YOUR_API_KEY.
Remote:
claude mcp add --scope user --header "Authorization: Bearer YOUR_API_KEY" \
--transport http context7 https://mcp.context7.com/mcp
Local:
claude mcp add --scope user context7 -- npx -y @upstash/context7-mcp --api-key YOUR_API_KEY
For the remote server, open Settings > Connectors > Add Custom Connector, name it Context7 and enter https://mcp.context7.com/mcp. For a local server, add this to claude_desktop_config.json:
{
"mcpServers": {
"context7": {
"command": "npx",
"args": ["-y", "@upstash/context7-mcp", "--api-key", "YOUR_API_KEY"]
}
}
}
Add to ~/.cursor/mcp.json (the README also has a one-click install button):
{
"mcpServers": {
"context7": {
"url": "https://mcp.context7.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
Add to .vscode/mcp.json:
{
"servers": {
"context7": {
"type": "http",
"url": "https://mcp.context7.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
Add to ~/.codex/config.toml:
[mcp_servers.context7]
url = "https://mcp.context7.com/mcp"
http_headers = { "Authorization" = "Bearer YOUR_API_KEY" }
Instructions for 30+ other clients are in the Context7 client docs.
/supabase/supabase or /vercel/next.js skip the search step.ctx7 library, ctx7 docs) for agents where you would rather not run MCP.npx ctx7 setup for Cursor, Claude Code and OpenCode.The server keeps its tool surface small, with two tools that are usually called in sequence:
resolve-library-id turns a plain library name ("supabase", "next.js") into a Context7-compatible ID. It takes libraryName and your query, which is used to rank matches.query-docs returns documentation and examples for a given libraryId and query. An optional language parameter ranks code examples in your language first.If you already know the library ID, put it in your prompt and the agent can go straight to query-docs.
/login. use context7"CLAUDE.md.@upstash/context7-mcp package still calls the Context7 API, so it needs network access either way.Authorization: Bearer (remote) or --api-key (local).ctx7sk, and that you are using the right header or flag for your transport./owner/repo or /owner/repo/version. Run resolve-library-id first and confirm the library exists on context7.com.node --version, try @upstash/context7-mcp@latest, or use bunx instead of npx for ERR_MODULE_NOT_FOUND. Switching to the remote URL avoids Node.js issues entirely.claude mcp list to verify the server, check claude mcp logs context7, or remove and add it again. Connectivity can be tested with curl https://mcp.context7.com/ping.Yes. Context7 and its MCP server are built by Upstash; the source is at github.com/upstash/context7 and the npm package is `@upstash/context7-mcp`.
There is a free plan. At the time of writing, [context7.com/plans](https://context7.com/plans) lists Free at $0 with 1,000 API calls per month and Pro at $10 per seat per month, plus Enterprise.
A key is recommended rather than strictly required; it gives you higher rate limits. Get one free at context7.com/dashboard, or let `npx ctx7 setup` generate it for you.
`https://mcp.context7.com/mcp`, used with Streamable HTTP and an `Authorization: Bearer YOUR_API_KEY` header.
Yes. All three are documented, along with Claude Desktop, Codex and more than 30 other clients.
`ctx7 setup` installs a skill that triggers it for library questions. Otherwise add a rule in Cursor Rules or `CLAUDE.md`, or end prompts with "use context7".
Yes. Library owners can submit projects; see the Adding Libraries guide in the Context7 docs.
Context7 says its projects are community-contributed and it cannot guarantee the accuracy, completeness or security of all documentation. Use the "Report" button on a project page if something looks wrong.

Search saved employee handbooks from your AI assistant. Four read-only tools list authorized handbooks, browse policies, search content, and read source pages with citation links. Uses scoped bearer tokens and Streamable
Groundwork: company memory for your AI tools. Docs public. Source closed. Public proof needs no account.

Shared memory for AI coding agents. Claude Code, Cursor and Codex read your project's decisions, lessons and tasks before they act.

Exact, shared memory for AI agents: a hierarchical database where totals always add up, every write is attributed to the agent and person behind it, and any figure traces back to its source file. Agents create databases,
Memlord is a remote MCP server available at https://app.memlord.com.
Comments