Open Source & MCP Integration

Docsieve is built to fit into modern agentic workflows. Run the parser locally, host the worker queue on your own infra, or query briefs directly inside your AI IDE via Model Context Protocol (MCP).

The Open Source CLI

You can execute local crawls, parse HTML trees, and compile briefs on your own machine using the `docsieve` Python package.

Installation

pip install docsieve

Basic Command Usage

To start a documentation recursive compile run and save the Markdown summary locally:

docsieve crawl https://docs.fastapi.tiangolo.com --depth 3 --output ./fastapi-brief.md

Model Context Protocol (MCP) Server

Model Context Protocol is an open standard that allows LLM clients (like Cursor or Claude Code) to securely interact with local databases, tools, and context. Docsieve provides a built-in MCP server to fetch and search briefs.

Starting the MCP Server

If you have `docsieve` installed, execute the following command to start the stdio server:

python -m docsieve_hosted.mcp

AI Editor Registration

{
  "mcpServers": {
    "docsieve": {
      "command": "python",
      "args": ["-m", "docsieve_hosted.mcp"]
    }
  }
}

Self-Hosting the Platform

Want to run the complete workspace dashboard and worker queue internally? Docsieve can be run via Docker Compose, hooking into your own Postgres and storage buckets.

Docker Compose Configuration

version: '3.8'

services:
  db:
    image: postgres:15
    environment:
      POSTGRES_DB: docsieve
      POSTGRES_PASSWORD: secret_password

  worker:
    image: docsieve-worker:latest
    environment:
      DATABASE_URL: postgresql://postgres:secret_password@db/docsieve
      REDIS_URL: redis://redis:6379/0

  api:
    image: docsieve-api:latest
    ports:
      - "8000:8000"
    environment:
      DATABASE_URL: postgresql://postgres:secret_password@db/docsieve