Agent-MCP multi-agent framework

Agent-MCP is a tool for people who want more than one AI helper working on a coding project at the same time. It lets you set up several AI agents, give each one a job, and have them share what they learn. It is aimed at experienced developers, not beginners.

What this MCP server does

You talk to your AI as usual and ask it to set up agents or assign work. Your AI passes that request to Agent-MCP, which runs in the background on your computer. Agent-MCP creates the agents, gives them tasks, and keeps a shared memory of the project so nothing gets lost. When you ask a question, it searches that memory and sends an answer back. You see the results in your chat, and can also open a dashboard to watch the agents work.

Click to zoom

What you can do with it

• Create specialized helper agents for backend, frontend, testing, or other roles
• Assign tasks to specific agents and track their progress
• Store and search your project's knowledge in a shared memory
• Send messages between agents or broadcast updates to all of them
• See a live dashboard of which agents are active and what they are doing
• Query the project memory to recall past decisions and patterns

Try asking your AI

• “Create a backend agent that specializes in API development”
• “Assign the task of adding user login endpoints to the backend agent”
• “What is our current database schema?”
• “List all active agents and what they are working on”

What it gives back to you

You get answers in your chat: lists of agents and their status, task progress, and text pulled from the project memory. When you ask a question, it replies with the matching information it found. When you create or assign something, it confirms what it did. The dashboard shows the same information as a live picture.

Before you start

Who it's for

Experienced developers and technical teams who already use AI coding assistants and want to coordinate several agents on a larger codebase.