Context7
Feeds your agent current, version-specific library docs instead of whatever it half-remembers from training. Fixes a real failure mode — by routing your questions through someone else's service.
What it actually does
Context7 answers the oldest complaint about coding with a model: the API it confidently describes shipped two majors ago. Ask for a library and it pulls current, version-specific documentation and working code examples from Upstash's index, injected into context right when the agent needs them. In practice you add one server and write 'use context7' in a prompt; the difference between guessed and quoted APIs shows up immediately. The trade is architectural: it's a hosted service, so your library queries flow through Upstash's API, and coverage is only as good as their index. For public open-source libraries that's a fine deal. For anything proprietary, it has nothing to offer by design.
Install
Read the source before you run this. Yes, actually.
From the desk
The Sunnyvale take
One of the few MCP servers that earns its install in the first ten minutes. The hosted dependency is the honest price: you're trading a hallucination problem for a availability-and-privacy question, and for public libraries that trade is worth taking.
Why it matters
Stale API knowledge is the most common way coding agents quietly fail. This is currently the lowest-friction fix, and its popularity has made 'docs lookup' a standard slot in the agent stack.
In practice
Pin a project to Next.js 16, tell the agent to 'use context7', and watch it stop suggesting App Router APIs that were renamed a year ago.
The skeptic's note
It solves training staleness for exactly the libraries popular enough to be indexed — which are also the ones models know best. The marginal win is real but biggest on the long tail, where coverage is thinnest.
Use cases
- Current API signatures while coding against fast-moving libraries
- Version-specific migration help
- Grounding library claims with quoted docs instead of memory
Risks & caveats
Read-only from your machine's perspective — no file or credential access. Your documentation queries are sent to Upstash's hosted API, so treat prompts around it as leaving the machine.
Related tools
Full indexSame shelf, different trade-offs.
Fetch MCP Server
The reference web-fetch server: give an agent a URL, get back the page as readable markdown. Small, honest, and easier to misuse than it looks.
GitHub MCP Server
GitHub's official MCP server — lets an agent read repos, file issues, and push code with your credentials, which is exactly as useful and as dangerous as it sounds.
Git MCP Server
The reference server for local git — status, diff, log, commit, and branch operations on repositories you point it at. Your history becomes agent-writable, which is most of the point and all of the risk.