MCP Server
Expert exposes a Model Context Protocol (MCP) server so AI agents can access your video knowledge as native tools. Available on the Enterprise plan.
What is MCP
The Model Context Protocol is an open standard for connecting AI agents to external tools and data sources. Instead of writing custom API integration code, MCP-compatible agents can discover and call Expert’s tools automatically, search videos, get video details, and list your library.
Endpoint and auth
MCP endpoint
https://expert.demobites.com/api/mcpAuthentication
All requests require a Bearer token in the Authorization header:
Authorization: Bearer YOUR_API_KEYAvailable tools
The MCP server exposes three tools that agents can discover and call:
search_videos
Search your video knowledge base with a natural language query. Returns matching segments with transcripts, timestamps, and confidence scores.
get_video
Retrieve full details for a specific video by ID, including the complete transcript and all caption segments.
list_videos
List all videos in your Expert with pagination support. Useful for browsing the full library.
For full request/response schemas and parameters, see the REST API reference. The MCP tools mirror the REST endpoints exactly.
Claude Desktop config
Add Expert to your Claude Desktop MCP configuration:
{
"mcpServers": {
"demobites-expert": {
"url": "https://expert.demobites.com/api/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Claude Code config
For Claude Code, add Expert to your project’s MCP settings:
{
"mcpServers": {
"demobites-expert": {
"type": "url",
"url": "https://expert.demobites.com/api/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Stateless design
Expert’s MCP server is stateless. Each request is independent, there are no sessions, no conversation history, and no state to manage. Your agent handles conversation context; Expert handles video knowledge retrieval.
No session management needed
Simply point your agent at the MCP endpoint with a valid API key. There is nothing to initialize, no handshake, and no cleanup. Every tool call is self-contained.