anaksor-ocr
Sokheng · Kit
Reads Khmer and English text from documents, receipts and photos of signs, using a detector for text lines and a recogniser for each line. Suits developers who…

Upstash’s hosted MCP server that hands your coding assistant current docs and code examples for the library and version you ask about.
Context7 is an MCP server for AI apps in the Developer category, listed by Mindket. It is free. This is version 1, last updated on . Add it to Claude Code with the line on this page, or put its JSON in Claude Desktop, Cursor or another app. Nothing runs until you add it.
npx ctx7 setup
Context7 is Upstash’s service that pulls current, version-specific documentation and code examples into your coding agent’s context. Left to its training data, a model can suggest calls that were renamed or never existed. With Context7 it answers from the docs for the version you name.
It works in two modes:
| Mode | What setup adds | How your agent uses it |
|---|---|---|
| CLI and skills | A skill | It runs ctx7 commands, with no MCP at all |
| MCP, this listing | A server registered with your agent | It calls Context7’s tools on Upstash’s hosted server |
Developers who code with an AI agent against libraries that change often, such as web frameworks and SDKs. Teams picking up an unfamiliar library get the most from it, and so does anyone tired of fixing calls a model made up.
| Industry | Who uses it | How they use it | Example |
|---|---|---|---|
| Web development | Front-end and full-stack developers on fast-moving frameworks such as Next.js | Ask for code that matches the framework version they run | A developer upgrading a framework asks how the new routing works and gets code for the version named. |
| Backend and SaaS | Developers wiring in an SDK such as Supabase’s | Set up an SDK they have not used before from its current docs | A developer asks for basic authentication with Supabase and names its library ID, so the matching step is skipped. |
| Agencies and freelancers | Developers who change stacks from one client to the next | Pull each client’s docs instead of trusting what the model remembers | A freelancer moving from a Next.js job to a Supabase one names the library and version in each question. |
| Software teams | Leads and reviewers who check code an agent wrote | Add a rule so agents use Context7 unasked, then trace a suggested call to its source | A reviewer asks which docs excerpt a new call came from and checks it before merging. |
Work mostly in a stable standard library and you will notice less.
It pays off when you write code against a library that moves fast, or set up an SDK you have not used before.
Setup takes one npx line.
| Need | Detail |
|---|---|
| Node.js | 22.13 or newer, for the setup command |
| A Context7 API key | Free from your Context7 dashboard. Upstash recommends one for higher rate limits. |
| A coding agent | Cursor, Claude Code, opencode or another MCP client |
npx ctx7 setup
To take the setup out again:
npx ctx7 remove
Add use context7 to a coding question, and put the version in the question when it matters. If you know the library’s Context7 ID, name it and the matching step is skipped.
How do I set up a Next.js 16 proxy? use context7
Implement basic authentication with Supabase. use library /supabase/supabase for API and docs.
| Tool | What it takes | What it returns |
|---|---|---|
| resolve-library-id | Your question and a library name | Matching Context7 IDs |
| query-docs | An ID such as /vercel/next.js and your question, plus an optional language | Docs and code examples, with examples in that language ranked first |
Setup also installs a skill that calls Context7 for library questions. Without it, a rule in CLAUDE.md or in Cursor’s rules does the same. The README’s example rule:
Always use Context7 when I need library/API documentation, code generation, setup or configuration steps without me having to explicitly ask.
One of the README’s own prompts, about caching on Cloudflare.
Configure a Cloudflare Worker script to cache JSON API responses for five minutes. use context7
The agent calls resolve-library-id with the library’s name and your question, picks a match, then calls query-docs with that ID. Its answer is built on the doc excerpts that came back. They stay in the chat, so you can ask which part a line of code came from.
If you are new. AI models learn from data with a cutoff date, so they may not know a library’s latest version. Context7 fetches the current docs and hands them to your agent before it answers. Add use context7 to your next coding question. Then compare the answer with one asked without it.
For experts. Naming the ID skips the first call, one round trip fewer per question. The skills mode does the same from a terminal:
ctx7 library <name> <query> ctx7 docs <libraryId> <query>
A server on the web: your app connects to its address. Put your own value where a line says <Authorization>.
Run this once in a terminal:
claude mcp add --transport http context7 https://mcp.context7.com/mcp --header "Authorization: <Authorization>"
In Claude Desktop, open Settings → Connectors → Add custom connector, and paste this address:
https://mcp.context7.com/mcp
Add this to .cursor/mcp.json in your project, or to ~/.cursor/mcp.json for every project:
{
"mcpServers": {
"context7": {
"url": "https://mcp.context7.com/mcp",
"headers": {
"Authorization": "<Authorization>"
}
}
}
}Add this to .vscode/mcp.json in your project; Copilot in agent mode then has its tools:
{
"servers": {
"context7": {
"type": "http",
"url": "https://mcp.context7.com/mcp",
"headers": {
"Authorization": "<Authorization>"
}
}
}
}Add this to ~/.codex/config.toml; the Codex CLI and its IDE extension both read it:
[mcp_servers.context7]
url = "https://mcp.context7.com/mcp"
http_headers = { "Authorization" = "<Authorization>" }In Windsurf, open the Cascade panel’s … menu → Open MCP config file, and add this to ~/.codeium/windsurf/mcp_config.json:
{
"mcpServers": {
"context7": {
"serverUrl": "https://mcp.context7.com/mcp",
"headers": {
"Authorization": "<Authorization>"
}
}
}
}NeedsAuthorization
Toolsresolve-library-idquery-docs
Listed MCP servers in the same category that share the most GitHub topics with this one.
| Product | Stars | Activity | License | Last change |
|---|---|---|---|---|
| Context7 This page | 64,331 | Active | MIT | |
| REA: Reverse Engineer AnythingShares mcp | 457,771 | Active | MIT | |
| Playwright MCPShares mcp | 39,312 | Active | Apache-2.0 |
Posts and comments that name this project on Hacker News, Bluesky, X, Reddit, Facebook and YouTube, found by a weekly search and on Trendshift. Mindket doesn’t write or check them. Open one to read it where it was posted.
Sokheng · Kit
Reads Khmer and English text from documents, receipts and photos of signs, using a detector for text lines and a recogniser for each line. Suits developers who…
Qwen · Kit
Qwen2.5-7B-Instruct as GGUF files in eight quantization sizes, to chat with locally through llama.cpp. For people who want a capable 7.6B model on their own co…
ACE-Step · Kit
Generates music from text with a 4B diffusion decoder, 50 steps and guidance control, aiming for higher audio quality than the base. For creators with a GPU of…
sansan0 · Kit
An open source news and public opinion monitor. It reads Chinese hot-search lists and your RSS feeds, keeps what matches your keywords or an AI score, and push…
cplj · Kit
Checks portfolio risk for your holdings from an AI chat: concentration, volatility and beta to SPY, plus scenarios under stated assumptions. A hosted MCP serve…
affaan-m · Skill
Affaan Mustafa’s plugin of agents, skills, commands and hooks that gives Claude Code, Codex and other coding agents a routine of planning, test-first work, fre…