autoteam

AutoTeam is an MCP hub that orchestrates AI agents across various services and platforms. It enables you to connect AI agents (such as Claude Code, Gemini CLI, Qwen Code, and other MCP-enabled tools) to external systems through a centralized MCP server. By running the AutoTeam MCP server in a container, you expose a universal integration point where agents can collaborate, share state, and execute multi-service workflows without bespoke integrations. To use it, install the AutoTeam MCP server (via Docker in this configuration), then configure MCP endpoints, agents, and service MCPs in your MCP network. Once running, agents can be assigned to tasks that involve multiple platforms (GitHub, Slack, databases, CMSs, etc.) and AutoTeam will coordinate parallel execution, routing work to the appropriate MCPs and agents as defined by your workflows.

Prerequisites:

• Docker installed and running on your host
• Network access to pull Docker images from Docker Hub

Installation steps:

1. Pull and run the AutoTeam MCP server as a container: docker run -i diazoxide/autoteam

2. If you want to customize configuration, create a local MCP network configuration file and mount it into the container as needed (depending on the image's supported config method).

3. Verify the server is up by checking container logs: docker ps docker logs <container_id>

4. Integrate with your MCP ecosystem by registering MCP endpoints for GitHub MCP, Slack MCP, Database MCP, etc., using the AutoTeam documentation under docs/configuration.md for guidance.

Note: If you prefer to build locally, you would need Go tooling and build the project from source, then run the resulting binary. The Docker approach is recommended for simplicity and isolation.

Tips and considerations:

• Environment variables: If you need to customize behavior (endpoints, tokens, timeouts), use environment variables supported by the image (refer to the image docs). The default setup assumes standard MCP endpoints are reachable from the container.
• Networking: Ensure the container can reach external MCP services (GitHub MCP, Slack MCP, etc.) and any target platforms.
• Scaling: AutoTeam is designed to orchestrate across multiple agents; consider running additional containers or a cluster setup to handle higher workloads.
• Troubleshooting: Check container logs for errors related to agent registration, MCP discovery, or network access. Look for missing token errors or refused connections, which typically indicate misconfigured endpoints or permissions.
• Security: Use least-privilege tokens for MCP integrations and secure your container runtime following best practices.