mcp-agent

The MCP Agent Server is an open-source backbone designed to empower AI employees and agents to operate alongside humans within a modular, brain-like system. It exposes endpoints and tooling that enable you to register users, manage API keys, issue and revoke access, and orchestrate AI agents with persistent memory, onboarding feedback loops, and agent-centric workflows. When running, the server serves a set of endpoints for creating and managing users, generating and listing API keys per user, and controlling agent behavior through the included powerShell-like commands and example endpoints. The design emphasizes proactive, adaptive, and personalized agent actions, enabling agents to learn from user feedback and memories to improve performance over time.

To use the server, start it with the configured command and access the endpoints documented in the API Auth section. Endpoints under /users, /users/me, /users/me/api-keys, and /health are available, with /health kept simple and fast. The API Key management system allows per-user keys that protect operations like registration, login, and key revocation. You can interact with key management through the provided HTTP methods (POST for registration and login, GET for listing keys, POST for creating keys, and DELETE for revoking keys). The server is engineered to support a robust, agent-centric workflow while remaining accessible to SMBs and individuals who need a straightforward, non-technical user experience for engaging with AI agents.

Prerequisites:

• Node.js installed on your machine
• Optional: Docker for containerized deployment

Step-by-step:

1. Clone the repository: git clone https://github.com/yourusername/mcp-agent-server.git cd mcp-agent-server

2. Remove any pre-existing comments or JS config if needed (per readme guidance).

3. Install dependencies: npm install

4. Run using Docker Compose (recommended for easy deployment): docker-compose up --build

5. Start the server directly (alternative): node path/to/server.js

6. Verify the server is running by visiting http://localhost:4000 (as indicated in the readme).

Prerequisites for Docker-based deployment:

• Install Docker and Docker Compose
• Ensure ports 4000 (and any other needed ports) are accessible

Tips:

• If you modify environment variables, restart the server to apply changes
• For production, consider using Docker Compose with a reverse proxy and persistent storage

Notes and tips:

• This MCP server discusses API Key management per user and agent-centric workflows, including memory persistence and feedback loops.
• Ensure you have your API keys ready if you plan to enable x-api-key authenticated endpoints.
• If endpoints return 404 or 401, verify that API keys are correctly configured and included in the request headers.
• For deployment, use the provided Docker setup to simplify environment configuration and portability.
• The system supports persistent memory and agent-centric features; to leverage this, consider following the full design outlined in the memory, persistence, and agent-life sections of the repository documentation.