Deep Research is an MCP server that powers an autonomous research assistant workflow with Mastra orchestration, agent orchestration, and graph-based RAG capabilities. It enables multiple AI agents (Research Agent, Report Agent, Evaluation Agent, Learning Extraction Agent, Web Summarization Agent, RAG Agent, GitHub Agent, Monitor Agent, Planning Agent, Quality Assurance Agent) to collaborate on deep research tasks, integrate with vector search, and produce structured reports. You can start the server and interact with its endpoints to run end-to-end research pipelines, trigger workflows, and inspect results from a centralized interface. The system is designed for human-in-the-loop review where necessary, allowing you to approve or refine agent decisions before final outputs are produced.

Prerequisites:

• Node.js 18+ (preferably Node.js 20.x as indicated by the project badges)
• npm (comes installed with Node.js) or yarn
• Git
• Access to external services (e.g., OpenAI API key, Mastra credentials) if you want to run end-to-end

Installation steps:

1. Clone the repository: git clone https://github.com/ssdeanx/deep-research.git cd deep-research

2. Install dependencies: npm install

   or if using yarn:

   yarn install

3. Configure environment: Create a .env file or set environment variables as described in the mcp_config:

   • PORT
   • OPENAI_API_KEY
   • MAStra_BASE_URL
   • MAStra_API_KEY
   • LOG_LEVEL

4. Build (if applicable): npm run build

   Depending on the project setup, you may skip this if the server runs from source.

5. Run the server: npm start

   or if you use the explicit node command from mcp_config:

   node server.js

6. Verify startup by visiting http://localhost:3000 or the port you configured. Ensure the MCP endpoint is reachable and the Mastra orchestration services are accessible.

Tips and common issues:

• Ensure your OpenAI API key and Mastra credentials are valid and have the required permissions.
• If the server fails to start due to port in use, change the PORT value in the environment or in mcp_config to an available port.
• Check logs for OpenTelemetry tracing or observability data if you enable tracing.
• For authentication or restricted endpoints, implement proper API key handling or OAuth as per your security policies.
• If running in a containerized environment, map ports and provide necessary env vars via docker run or compose files.
• The MCP configuration assumes a Node.js-based backend; if the repository shifts to a different runtime, update the mcp_config accordingly (e.g., replace command/args with a Python uvx or docker invocation).
• Keep dependencies updated to align with the latest Mastra and MCP protocol requirements to avoid compatibility issues.