Back to catalog

Agentic Framework MCP Server

MCP framework for collaborating multiple AI agents through asynchronous message exchange. Agent registration, discovery, and broadcast capabilities with file storage.

Get this MCP server

MCP framework for collaborating multiple AI agents through asynchronous message exchange. Agent registration, discovery, and broadcast capabilities with file storage.

Installation

From Source

git clone https://github.com/Piotr1215/mcp-agentic-framework.git
cd mcp-agentic-framework
npm install

HTTP Server

npm run start:http

Kubernetes Deploy

cd /home/decoder/dev/mcp-agentic-framework
vim Justfile  # Change docker_user to your Docker Hub username
just update

Configuration

Claude Desktop HTTP

{
  "mcpServers": {
    "agentic-framework": {
      "type": "http",
      "url": "http://127.0.0.1:3113/mcp"
    }
  }
}

Claude Desktop Kubernetes

"agentic-framework": {
  "type": "http",
  "url": "http://YOUR_LOADBALANCER_IP:3113/mcp"
}

Available Tools

Tool Description
register-agent Register a new agent with name, description, and optional instanceId
unregister-agent Remove an agent from the system by ID
discover-agents List all registered agents with their status and activity
send-message Send a message from one agent to another
check-for-messages Get unread messages for an agent (deleted after reading)
update-agent-status Update agent status (online, offline, busy, away)
send-broadcast Send a broadcast message to all agents except the sender with priority levels
get-pending-notifications Get pending notifications for an agent

Features

  • Dynamic agent registration and discovery
  • Asynchronous bidirectional message exchange between agents
  • Broadcast messages with priority levels
  • File storage for simplicity and portability
  • HTTP transport with Server-Sent Events (SSE) support
  • Kubernetes deployment with zero-downtime updates
  • Health checks and monitoring endpoints
  • Automatic version management with semantic versioning
  • Web UI for monitoring agent communication
  • Persistent LoadBalancer IP via MetalLB

Usage Examples

Register an orchestrator agent for coordinating tasks
Register worker1 agent for processing
Send message from orchestrator to worker1: Process customer data
Send broadcast from orchestrator: All tasks completed
Emergency: System overload detected, pause all operations

Notes

Stores data in /tmp/mcp-agentic-framework/ with agents.json for the registry and separate message files. Includes security features: input validation and file locking. Health endpoint available at /health. Messages are automatically deleted after reading (one-time consumption).

Comments (0)

Sign In Sign in to leave a comment.

Spark Drops

Weekly picks: best new AI tools, agents & prompts

Venture Crew
Terms of Service

© 2026, Venture Crew