Context7

Context7

Context7 is an MCP server that feeds AI coding assistants up-to-date, version-specific library documentation. Context7 pulls current docs and code examples straight from the source into the model's context, which cuts the hallucinated APIs and outdated snippets that LLMs produce when they rely on stale training data. Built by Upstash, Context7 runs as a remote Model Context Protocol server or a local npm package and integrates with 30+ clients including Cursor, Claude, VS Code, and Windsurf. The server is open source under the MIT license.

Production credibility: A product of Upstash, Inc. (the serverless data platform for Redis and Kafka, founded 2021 in San Jose; CEO Enes Akar). Upstash has raised roughly $11.9M total from backers including Andreessen Horowitz and Bek Ventures. The open-source Context7 repository has around 57,000 GitHub stars and is MIT-licensed; the docs-parsing backend is proprietary.

Key Features

• Up-to-date, version-specific library docs and code examples injected into model context
• MCP tools resolve-library-id and query-docs
• Remote Streamable-HTTP endpoint plus a local npm/stdio server
• Integrations with 30+ MCP clients (Cursor, Claude, VS Code, Windsurf, OpenCode)
• Optional free API key for higher rate limits
• CLI and Skills mode that works without MCP
• Open-source server under the MIT license

Ideal Use Case

Developers using AI coding assistants who keep getting outdated or invented API calls — Context7 puts the current, version-correct docs for the libraries they use directly into the model's context, so generated code matches the version they are actually on.

How Context7 differentiates

General MCP registries help you find tools; Context7 is a single, focused server that solves one problem well — feeding accurate, version-specific documentation into the model. It works across 30+ clients, offers a no-MCP CLI mode for setups that cannot run a server, and is open source under MIT with the backing of Upstash. For developers whose pain is hallucinated or stale API code, it is the documentation-grounding server rather than a directory.

FAQ

Q: What is Context7? A: Context7 is an MCP server by Upstash that feeds AI coding assistants up-to-date, version-specific library documentation and code examples, reducing hallucinated or outdated API calls.

Q: How does Context7 reduce hallucinated code? A: It pulls current docs straight from the source into the model's context at query time, so the model works from accurate, version-correct documentation instead of stale training data.

Q: Which editors support Context7? A: Context7 integrates with 30+ MCP clients including Cursor, Claude, VS Code, Windsurf, and OpenCode, via a remote endpoint or a local npm server.

Q: Is Context7 free and open source? A: The Context7 MCP server is open source under the MIT license and free to use, with an optional API key for higher rate limits and paid tiers for more usage. The docs-parsing backend is proprietary.

Q: Do I need MCP to use Context7? A: No — besides the MCP server, Context7 offers a CLI and a Skills mode that work without MCP, for setups that cannot run a server.

tl;dr

Context7 is an Upstash MCP server that injects up-to-date, version-specific library docs into AI coding assistants, cutting hallucinated and outdated API calls. Remote endpoint or local npm server, 30+ clients, MIT-licensed (~57K GitHub stars). The documentation-grounding pick for AI pair programming.