Docsie

Created By

Philippe10 days ago

Docsie MCP brings your documentation workflow directly into Claude. Create, edit, and publish docs without leaving your AI assistant. Convert videos into structured documentation instantly, run compliance analysis on text, audio, and video content, and use semantic search to find exactly what you need across your knowledge base. Whether you're building product docs, SOPs, or technical guides, Docsie MCP makes it faster and smarter — all powered by AI.

# documentation

# productivity

Overview Content Tools Comments

Overview

Docsie Video-to-Docs MCP Server

Remote MCP server for Anthropic's Connectors Directory. Converts videos into structured documentation using Dokuta AI.

What It Does

Users on Claude can:

Submit a video URL (YouTube, Loom, Vimeo, direct MP4, etc.)
Choose a doc type (user guide, SOP, product docs, policy, blog)
Get structured markdown documentation back

Free tier: Videos up to 5 minutes are processed free. Longer videos require a Docsie account with AI credits.

Architecture

Claude → POST https://mcp.docsie.io/mcp (Bearer token)
              │
              ├─→ Docsie /o2/ (OAuth2 token validation)
              ├─→ Dokuta API (video analysis)
              └─→ Docsie /api/internal/mcp/ (credit deduction)

Transport: Streamable HTTP (MCP spec)
Auth: OAuth2 Authorization Code + PKCE via Docsie's /o2/ provider
Runtime: Python 3.12 + FastMCP + Starlette + uvicorn

MCP Tools

Tool	Description
`analyze_video`	Submit a video URL for documentation generation
`check_job_status`	Poll for progress and retrieve completed results
`list_doc_types`	List available doc types and quality tiers

Local Development

cd mcp-server

# Install dependencies
pip install -e .

# Configure environment
cp .env.example .env
# Edit .env with your Dokuta API key and Docsie internal URL

# Run
uvicorn app.main:app --reload --port 8000

Test endpoints

# Health check
curl http://localhost:8000/health

# OAuth metadata
curl http://localhost:8000/.well-known/oauth-authorization-server

# MCP tool listing (via MCP protocol)
curl -X POST http://localhost:8000/mcp \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":"1","method":"tools/list","params":{}}'

Environment Variables

Variable	Description	Default
`MCP_DOCSIE_BASE_URL`	Docsie public URL (for OAuth metadata)	`https://app.docsie.io`
`MCP_DOCSIE_INTERNAL_URL`	Docsie internal URL (for API calls)	`http://docsie-web:8080`
`MCP_DOKUTA_API_URL`	Dokuta video analysis API	`https://app.videodokuta.com/api/v1`
`MCP_DOKUTA_API_KEY`	Dokuta API key	(required)
`MCP_MCP_INTERNAL_API_KEY`	Shared secret for Django internal endpoints	(required)
`MCP_OAUTH2_CLIENT_ID`	OAuth2 client ID for the MCP app	(required)
`MCP_FREE_TIER_MAX_SECONDS`	Max free video duration in seconds	`300`

Deployment (Kubernetes)

Docker images are built and pushed automatically via GitHub Actions on push to main.

# Deploy with Helm (secrets managed via Doppler)
helm upgrade --install mcp-staging helm/docsie-mcp/ \
  -n mcp \
  --set image.tag=v2026.04.06.01 \
  --set doppler.token=$DOPPLER_SERVICE_TOKEN

Django Side Setup

1. Add internal API key to Django settings

In your .env or environment:

MCP_INTERNAL_API_KEY=your-shared-secret-here

2. Register OAuth2 Application

In Django admin at /o2/applications/register/:

Name: Claude MCP Connector
Client type: Public
Authorization grant type: Authorization code
Redirect URIs: (provided by Anthropic during directory registration)
Skip authorization: Yes

3. Internal endpoints (already wired)

GET /api/internal/mcp/user-context/ — token validation + user info
POST /api/internal/mcp/deduct-credits/ — credit deduction

Anthropic Connectors Directory Submission

To submit to the directory, you'll need:

MCP Server URL: https://mcp.docsie.io/mcp
OAuth2 metadata URL: https://mcp.docsie.io/.well-known/oauth-authorization-server
Company name: Docsie Inc.
Description: "Convert videos into structured documentation using AI. Submit a video URL and get back a user guide, SOP, product docs, or blog post."
Logo: Docsie logo (SVG/PNG)
Privacy policy: https://www.docsie.io/privacy-policy/
Terms of service: https://www.docsie.io/terms-of-service/

Submit at: https://console.anthropic.com/dashboard/connectors

Server Config

{
  "mcpServers": {
    "docsie": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://app.docsie.io/mcp"
      ],
      "env": {
        "API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

Recommend Servers

TraeBuild with Free GPT-4.1 & Claude 3.7. Fully MCP-Ready.

CursorThe AI Code Editor

MCP AdvisorMCP Advisor & Installation - Use the right MCP server for your needs

MiniMax MCPOfficial MiniMax Model Context Protocol (MCP) server that enables interaction with powerful Text to Speech, image generation and video generation APIs.

ChatWiseThe second fastest AI chatbot™

Zhipu Web SearchZhipu Web Search MCP Server is a search engine specifically designed for large models. It integrates four search engines, allowing users to flexibly compare and switch between them. Building upon the web crawling and ranking capabilities of traditional search engines, it enhances intent recognition capabilities, returning results more suitable for large model processing (such as webpage titles, URLs, summaries, site names, site icons, etc.). This helps AI applications achieve "dynamic knowledge acquisition" and "precise scenario adaptation" capabilities.

Jina AI MCP ToolsA Model Context Protocol (MCP) server that integrates with Jina AI Search Foundation APIs.

Amap Maps高德地图官方 MCP Server

WindsurfThe new purpose-built IDE to harness magic

AiimagemultistyleA Model Context Protocol (MCP) server for image generation and manipulation using fal.ai's Stable Diffusion model.

Baidu Map百度地图核心API现已全面兼容MCP协议，是国内首家兼容MCP协议的地图服务商。