// the find
SolaceLabs/solace-agent-mesh
An event-driven framework designed to build and orchestrate multi-agent AI systems. It enables seamless integration of AI agents with real-world data sources and systems, facilitating complex, multi-step workflows.
Solace Agent Mesh is a Python framework for wiring together multiple AI agents over Solace's event broker, using Google's ADK for the agent runtime and the A2A protocol for inter-agent communication. It is aimed at enterprise teams who already have or are willing to run Solace infrastructure and want genuinely async, decoupled agent coordination rather than orchestration glued together with function calls.
The event-driven transport is the real differentiator here — agents communicate through an actual message broker, not a shared in-process call stack, which means you can scale and deploy agents independently without them knowing about each other. The layering on top of Google ADK is smart: they inherit a real agent runtime with tool execution and state management instead of rolling their own. The CLI tooling is solid — 'sam add agent --gui' and 'sam plugin install' lower the floor for getting a new agent wired in without touching YAML by hand. MCP integration works out of the box, so if your tools are already wrapped as MCP servers you don't have to rewrite them.
The Solace broker dependency is load-bearing and the README buries it. Getting a Solace PubSub+ instance running — even the free Docker image — is not a five-minute task, and the free tier limits will bite you before you reach anything resembling a real workload. The abstraction stack is deep: your code talks to SAC, which wraps ADK, which wraps the LLM provider, which all sit behind the Solace transport layer — when something breaks it's genuinely hard to know which layer is lying to you. The upgrade path is explicitly unsupported in the README ('not officially supported at this time'), which is a significant red flag for anything beyond a prototype. Windows requires WSL with no native support, which is unusual for a repo that claims enterprise readiness.