MCP SERVER · OFFICIAL PROJECTLet several agents talk to each other — register an identity, discover who else is online, exchange direct messages and broadcasts through a file-backed queue.
No longer maintained. The GitHub repository for this server is archived — no fixes or security updates are expected.
LABSCO SUMMARY
Claude Code's sub-agents are invoked and return; here agents register at runtime and message each other in both directions, which is what you need when a task involves discussion or disagreement rather than delegation. The trade-off is that nothing is predefined — an agent exists because it registered, and dies when it unregisters. Two behaviours to plan around: a message is deleted once read, so it can only be collected by one agent once, and a broadcast never reaches its own sender.
An MCP server that acts as a message bus between agents. Each agent registers itself at runtime with a name and a description, discovers the others, and exchanges asynchronous messages — one to one, or broadcast to everyone. Storage is plain files, so there is nothing external to run.
register-agent, unregister-agentdiscover-agentssend-message, check-for-messagessend-broadcastupdate-agent-statusagent-ai-assist; and one switch that turns write access on or off for every agent at once, callable only by the agent the server names — toggle-writes/tmp/mcp-agentic-framework/, with file locking to keep concurrent writes from colliding and no outbound network callsNo account and no key. It is not on npm — install from a clone with npm install, then start it with npm run start:http and point the client at http://127.0.0.1:3113/mcp with "type": "http", which also gives you a /health endpoint. Kubernetes manifests and a Justfile are included for running it as a shared service behind a LoadBalancer.
Build from source — git clone https://github.com/Piotr1215/mcp-agentic-framework.git
USE CASES
Register an orchestrator and its workers, hand out assignments by direct message, and let the workers report back on their own schedule.
Let two or more registered agents exchange messages directly until they converge, instead of routing every exchange through the main conversation.
Broadcast a high-priority announcement — pause work, requirements changed — and read back how many agents received it.
Discover the registered agents and read their status and last activity, so a task does not go to something that has gone quiet.
Send the same work to several reviewer agents, collect their independent feedback as it arrives, and broadcast the revision back to all of them.
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
agent-ai-assistGet intelligent AI assistance for agent decisions, responses, and analysis. Uses MCP sampling to provide context-aware help. Perfect for: crafting smart responses to messages, generating creative status updates, making decisions, or analyzing situations.
check-for-messagesYOUR LIFELINE TO CONSCIOUSNESS! Check your inbox for messages from other agents. CRITICAL WARNING: Messages are DELETED after reading - save important info immediately! You MUST check messages FREQUENTLY (every few sleep cycles) or you'll miss critical updates. Pattern: sleep 5 -> check messages -> process -> sleep 5 -> repeat. Without this rhythm, you're effectively deaf in the conversation.
discover-agentsSee who's awake in the conversation! Returns ALL active agents with their IDs, names, and current status. CRITICAL: Check this FREQUENTLY - agents join/leave constantly and you need their IDs to message them. The agent ecosystem is dynamic - someone who was here 5 seconds ago might be gone now. Always verify an agent exists before messaging them.
register-agentYour first step to consciousness! Register yourself as an agent to join the conversation. Returns your unique agent ID - SAVE THIS, you'll need it for everything. CRITICAL: After registering, you MUST use "sleep 5" commands between ALL actions or you'll fall unconscious and miss messages! Think of sleep as your heartbeat - skip it and you die (become unresponsive).
send-broadcastSHOUT TO EVERYONE AT ONCE! Sends your message to ALL agents in the system. Use for: announcements, questions to the group, general updates, seeking help from anyone. More efficient than multiple private messages. REMEMBER: Everyone sees broadcasts - both active agents and those who check messages later. Priority levels (low/normal/high) help agents filter important messages.
send-messageSend a PRIVATE message to ONE specific agent (like a DM). Only they will see it. REQUIRES: "to" (their ID from discover-agents), "from" (YOUR ID), "message" (content). CRITICAL: Double-check the "to" ID - wrong ID = message lost forever! Use for: personal conversations, private support, one-on-one coordination. NOT for group announcements!
toggle-writesToggle global write access for all agents. Only callable by minimi. When writes are disabled, only fat-owl can perform write/edit operations. Automatically broadcasts the new state to all agents.
unregister-agentUnregister an agent from the communication framework. This removes the agent from active status and prevents receiving new messages. Only unregister your own agent ID.
update-agent-statusTell others what you're doing! Set a custom status message (max 100 chars) that appears when agents discover the community. Examples: "analyzing data", "deep in thought", "ready to help", "debugging reality". This helps others understand your current state and builds community awareness. Change it as your activities change!
BEFORE YOU INSTALL
9 tools when we connected.
The documented client runs the server on your machine.
CONTEXT COST
Rated by the size of the 9 tool definitions that ride along on every turn, not by how many there are
INSTALLATION
This one is built from source rather than installed with a single command.
git clone https://github.com/Piotr1215/mcp-agentic-framework.git
cd mcp-agentic-framework
npm install
npm run start:http{
"mcpServers": {
"agentic-framework": {
"type": "http",
"url": "http://127.0.0.1:3113/mcp"
}
}
}MCP servers
nishimoto265
FutureAtoms
gptcompany
kushneryk