An open-source event-driven multi-agent AI framework built on Solace Event Mesh, enabling task orchestration and enterprise integration among specialized agents via the A2A protocol.
Overview#
Solace Agent Mesh (SAM) is an open-source framework maintained by SolaceLabs, designed to build applications where multiple specialized AI agents work together. It leverages the event messaging capabilities of the Solace Platform for true scalability and reliability.
Core Features#
- Event-Driven Architecture: Agent communication via Solace Event Mesh for low coupling and high concurrency
- A2A Protocol Support: Native Agent-to-Agent communication protocol for agent discovery and task delegation
- Orchestrator: Automatically decomposes complex tasks and assigns them to specialized agents
- Dynamic Embeds: Embed real-time data, computation results, and file content in responses
- Multi-Channel Access: Native support for REST API, Web UI, Slack, and Microsoft Teams
- Low-Code Extension: Add custom agents, gateways, or services with minimal code
Architecture#
Core architecture concept: "Universal A2A Agent Host" integrating these key components:
| Component | Version | Function |
|---|---|---|
| Google ADK | 1.18.0 | Agent runtime, LLM interaction, tool execution, state management |
| Solace AI Connector | 3.3.2 | Proxy connection, config loading, component lifecycle |
| A2A SDK | 0.3.7 | Client-agent and agent-agent communication |
| LiteLLM | 1.76.3 | Multi-LLM provider support |
Quick Start#
# Create project directory
mkdir my-sam && cd my-sam
python3 -m venv .venv && source .venv/bin/activate
# Install framework
pip3 install solace-agent-mesh
# Initialize project (GUI mode, port 5002)
sam init --gui
# Run project
sam run
# Access http://localhost:8000 for Web UI
Typical Use Cases#
| Scenario | Description | Est. Time |
|---|---|---|
| Weather Agent | Get real-time weather info | ~15 min |
| SQL Database Integration | Answer enterprise-specific questions | 10-15 min |
| MCP Integration | Integrate Model Context Protocol server | 10-15 min |
| Slack Integration | Chat with system from Slack | 20-30 min |
CLI Commands#
sam init --gui: Graphical project initializationsam run: Start agent mesh servicesam add agent --gui: Graphically add new agentsam plugin add <component> --plugin <name>: Install plugin
System Requirements#
- Python ≥ 3.10.16
- OS: MacOS, Linux, Windows (WSL)
- LLM API Key (mainstream providers supported)
Project Structure#
src/: Core source codecli/: CLI toolsclient/: Frontend client (with Web UI)config_portal/: Configuration portaldocs/: Documentation (Docusaurus)examples/: Examples (agents, gateways, services, workflows)templates/: Project templates