🚀 Introducing SBP: Multi-Agent Coordination via Digital "Pheromones

Naveen Thangavelu — Sun, 08 Feb 2026 16:29:51 +0000

We just open-sourced SBP (Stigmergic Blackboard Protocol), a protocol designed to let AI agents coordinate through environment-based signals rather than direct messaging or rigid orchestration.

The Problem: The "Orchestrator" Bottleneck

Most multi-agent systems rely on a central "brain" or complex message queues. This creates:

Single points of failure: If the orchestrator hangs, the system dies.
Brittle coupling: Adding a new agent often requires updating the logic of every other agent.
Noise: Direct messaging scales poorly as the number of agents increases.

The Approach: Stigmergy

SBP is inspired by stigmergy: the mechanism ants use to coordinate. Instead of Agent A telling Agent B what to do, agents leave "signals" on a shared blackboard.

Key Mechanics:

Signal Intensity: Agents post signals with varying strengths.
Decay Curves: Signals aren't permanent. They fade over time (or through interaction), preventing the "blackboard" from becoming cluttered with stale data.
Decoupled Awareness: Agents simply "sense" the blackboard and react based on their own internal logic.

SBP vs. MCP (Model Context Protocol)

We see these as complementary standards, not competitors:

MCP defines Capabilities: "What tools can I use and what data can I see?"
SBP defines Awareness: "What is the current state of the collective task and where is my help needed?"

🛠️ What’s Included

Full Protocol Specification: RFC 2119 compliant.
Reference Server: TypeScript implementation (@advicenxt/sbp-server).
SDKs: Ready-to-use libraries for TypeScript and Python.
Pluggable Storage: Ships with in-memory support; extensible to Redis or SQLite.
DevOps Ready: OpenAPI 3.1 spec and Docker support.

🔗 Links & Installation

GitHub: AdviceNXT/sbp
npm: npm install @advicenxt/sbp-server
PyPI: pip install sbp-client

Why we built this

We’re currently using SBP in production to manage non-linear workflows where agents need to "pick up where others left off" without a human or a script explicitly handing off the baton.

I'm happy to answer any questions about the protocol design, the math behind the decay mechanics, or how to implement a custom sensing loop!