From URL to AI-ready in minutes
No training pipelines. No complex configuration. Add a source and your AI gets perfect context.
Add a Source
Paste a URL, point to a GitHub repo, or drop in your docs. We find the pages, you don't have to list them.
Supports websites, sitemaps, GitHub repos
Auto-Index
We scrape, chunk, and embed everything in the background. Hybrid keyword plus vector search means results hold up in practice.
Runs automatically, re-syncs on schedule
Query via AI
Drop one config block into Claude Desktop or your editor. Ask a question, get the right docs back, every time.
MCP-native, zero extra tooling required
Built for AI-first developers
Fast Indexing
Paste a URL and walk away. Most documentation sites are ready to search in under five minutes.
MCP Native
Works out of the box with Claude Desktop, Cursor, Zed, and Windsurf. One config block and you're done.
Hybrid Search
Combines keyword matching with semantic vector search so results stay good whether you're specific or vague.
Pre-indexed Sources
React, Python, Vue, and more are already indexed. Enable them with one click, no waiting required.
Your Data Stays Yours
Internal documentation never leaves your workspace. Role-based access keeps the right content with the right people.
Open Source
Full source on GitHub under MIT. Run it yourself or use our hosted version. Either way works fine.
One config block.
Infinite context.
Drop this into your Claude Desktop config. That's it. Every indexed source becomes available to your AI straight away.
Get your API key{ "mcpServers": { "contextstream": { "command": "npx", "args": ["-y", "@contextstream/mcp"], "env": { "API_KEY": "cs_live_••••••••" } } } }