NOTE: This repo will be moved to https://github.com/upstash/context7-mcp
Instant LLM Context for AI Agents and Developers
Context7 is a robust platform designed to create, manage, and leverage context libraries for Large Language Models (LLMs), AI agents, and developers. It automatically pulls high-quality, targeted code snippets from documentation repositories, delivering them as ready-to-use context for AI systems. Built by the Upstash team, you can enjoy Context7 with confidence knowing it’s backed by proven expertise.
Key Benefits:
✅ Precise Context: Extracts clean, relevant snippets from up-to-date documentation
✅ Focused Content: Includes only code and descriptions—no fluff or filler
✅ Free Tier: Up to 50 queries per day for personal use at no cost
✅ Broad Compatibility: Works seamlessly with MCP servers (e.g., Cursor, Windsurf)
✅ Fast Integration: Generate library-specific context in minutes and embed Context7 links directly into your documentation
Context7 transforms documentation into actionable context through a streamlined pipeline:
-
Document Parsing: Supports a variety of formats:
- Markdown (
.md,.mdx) - Plain Text (
.txt) - ReStructuredText (
.rst) - Jupyter Notebooks (
.ipynb)
- Markdown (
-
Context Extraction: Uses LLM models to pull code snippets and craft concise, descriptive metadata.
-
Embedding Generation: Converts snippets and metadata into vector embeddings for fast, accurate retrieval.
-
Contextual Retrieval: Delivers relevant code examples instantly via API or web interface.
- Add your library to Context7 using one of the methods below.
- Once processed, access your documentation as contextual snippets.
- Refine results by searching topics and adjusting token size as needed.
Boost your docs with instant code examples by adding a Context7 link:
<a href="https://context7.com/YOUR_LIBRARY/llms.txt?tokenLimit=MAX_TOKEN_LIMIT" target="_blank">
See llms.txt with code examples
</a>This keeps users in your documentation while giving them quick access to implementation snippets.
Head to context7.com/add-library and add your library via our simple web interface.
- Create a JSON file with your library details (see schema below).
- Submit a Pull Request to our repository.
- Once approved and merged, your library will be indexed and available.
Important
Ensure the repo includes the documentation files of the library with md/mdx/txt/rst/ipynb format.
{
"settings": {
"library": "library-name", // Unique, URL-friendly identifier
"title": "Human-Readable Title", // Display name
"docsRepoUrl": "https://github.com/organization/repo",
"folders": ["docs"], // Optional: folders to include
"excludeFolders": ["archive", "old"] // Optional: folders to exclude
},
"version": {
"lastUpdate": "2023-03-25", // ISO 8601 timestamp
"totalTokens": 0, // Set by system after processing
"totalPages": 0, // Set by system after processing
"totalSnippets": 0, // Set by system after processing
"averageTokens": 0, // Set by system after processing
"parseDuration": 0, // Set by system after processing
"state": "initial", // Initial state for new libraries
"errorCount": 0 // Set by system after processing
}
}| State | Description |
|---|---|
initial |
Library created, not yet processed |
parsed |
Documentation parsed, not yet finalized |
finalized |
Fully processed and ready to use |
error |
Processing hit an error |
invalid_docs |
Docs couldn’t be parsed correctly |
stop |
Processing halted manually |
delete |
Library slated for removal |
Coming soon!
We’d love your input! See our contribution guidelines to get started.
Help us improve by reporting bugs or suggesting features here.