Yes. Dosu provides an MCP (Model Context Protocol) Server that lets AI coding assistants like Claude Code, Cursor, and VS Code Copilot search your Dosu knowledge base.

What the MCP Server Does#

The MCP Server exposes tools to connected AI assistants for researching questions across your organization's data sources:

Init knowledge: Search curated knowledge including documentation, approved answers, and topics. This is the starting point for research.

List available data sources: Lists all available data sources (GitHub repos, Notion pages, Slack channels, Confluence spaces, etc.) that can be queried. Returns data source IDs, names, providers, and descriptions.

Ask: Runs a comprehensive multi-step research agent across selected data sources to answer questions. Searches across codebases, documentation sources, and communication channels, then returns answers with citations, confidence scores, and observations. Ideal for research questions that require synthesizing information from multiple places.

Search documentation: Search raw documentation sources—Notion, Confluence, GitHub wikis, and web docs. Use when you need documentation specifically or when init knowledge doesn't find what you need.

The recommended workflow is: init_knowledge (for initial context) → list_available_data_sources (to discover searchable sources) → ask (to get comprehensive answers from selected sources).

Supported Clients#

Once enabled, you can connect the Dosu MCP Server to:

• Claude Code (via CLI)
• Cursor (via MCP config)
• Other MCP-compatible tools

Setup instructions are provided in the Dosu dashboard when you create an MCP deployment.