modus-agendi/managed-agent-control-mcp

What it does

Managed Agent Control MCP is a Model Context Protocol (MCP) server that provides comprehensive agent management capabilities for Claude Managed Agents. It allows users to programmatically start agents, observe their execution in real-time by polling events, interact with running agents, approve or deny tool executions, and gracefully stop agent operations from any MCP-compatible client.

How to set up

Install the MCP server from the modusagendi GitHub repository. The tool is available as a Python package and supports multiple platforms including macOS, Windows, Linux, and cloud environments. Configuration involves setting up your MCP client to connect to this server and providing appropriate Claude API credentials for agent management.

Key features

• Start and launch Claude Managed Agents programmatically
• Real-time event polling to observe agent execution and progress
• Interactive agent interaction and communication
• Tool execution approval workflow for controlled automation
• Agent lifecycle management with stop/terminate capabilities
• Cross-platform support (macOS, Windows, Linux, cloud)
• MCP-compatible for use with various client applications