Biel.ai

Biel.ai MCP Server Connect your IDE to your product docs

Give AI tools like Cursor, VS Code, and Claude Desktop access to your company's product knowledge through the Biel.ai platform.

Biel.ai provides a hosted Retrieval-Augmented Generation (RAG) layer that makes your documentation searchable and useful to AI tools. This enables smarter completions, accurate technical answers, and context-aware suggestions—directly in your IDE or chat environment.

When AI tools can read your product documentation, they become significantly more helpful—generating more accurate code completions, answering technical questions with context, and guiding developers with real-time product knowledge.

Note: Requires a Biel.ai account and project setup. Start your free 15-day trial.

See quickstart instructions →

Getting started

1. Get your MCP configuration

{
  "mcpServers": {
    "biel-ai": {
      "description": "Query your product's documentation, APIs, and knowledge base.",
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.biel.ai/sse?project_slug=YOUR_PROJECT_SLUG&domain=https://your-docs-domain.com"
      ]
    }
  }
}

Required: project_slug and domain
Optional: api_key (only needed for private projects)

2. Add to your AI tool

• Cursor: Settings → *Tools & Integrations → New MCP server.
• Claude Desktop: Edit claude_desktop_config.json
• VS Code: Install MCP extension.

3. Start asking questions

Can you check in biel_ai what the auth headers are for the /users endpoint?

Choose generated answers or search

The biel_ai tool accepts mode: "answer" | "search". Omit it to keep generated answers and conversation context. Use search with concise terms to retrieve ranked content from indexed web pages, uploaded files, repositories and OpenAPI sources without waiting for Biel.ai to generate an answer. The calling model can then compose an answer from those sources.

{"message": "SDK authentication", "mode": "search"}

Search requests the API's content_scope=all. Results include chunk text, source type and any page/sheet reference; non-URL documents are cited by title and location. The limit argument controls how many chunks are returned (default 5, maximum 20). Search does not create or continue a chat. Private projects require an API key with the search scope; generated answers use chats_create. Both paths retain the API's project access, domain and quota checks.

Self-hosting (Optional)

For advanced users who prefer to run their own MCP server instance:

Local development

# Clone and run locally
git clone https://github.com/TechDocsStudio/biel-mcp
cd biel-mcp
pip install .
biel-mcp

Use as a Python package

Applications that host their own ASGI stack can install the server directly from a tagged revision and import its FastAPI application:

pip install "biel-mcp @ git+https://github.com/TechDocsStudio/biel-mcp.git@VERSION"

from biel_mcp.server import app

Docker deployment

# Docker Compose (recommended)
docker-compose up -d --build

# Or Docker directly
docker build -t biel-mcp .
docker run -d -p 7832:7832 biel-mcp

Request timeout and continuation

BIEL_MCP_READ_TIMEOUT_SECONDS sets the upstream API read timeout (default: 60 seconds). Connect and pool timeouts remain 5 seconds, and writes have a 10-second timeout. Client deadlines can still end a request earlier.

Conversation credentials remain in the MCP session. After HTTP 404 for an expired session, initialize a new connection without the old session id. Tool failures return isError: true; the server logs mcp.upstream_completed with duration, status and the upstream request id.

Support

• Issues: GitHub Issues
• Contact: support@biel.ai
• Custom Demo: Book a demo