SolaceProducts/solace-agent-mesh

Platform for building and running AI agents across the full lifecycle — build, test, deploy, monitor, improve. Hosts your agents, runs their tools, and coordinates them over an event mesh, with sessions, artifacts, auth, evaluators, and observability built in.

What it is

Solace Agent Mesh is a platform for building and running AI agents — software programs that perform tasks on your behalf. It covers the whole lifecycle: you build an agent, test it, deploy it, monitor it, and improve it. The platform hosts the agents you create, runs the tools those agents call, and coordinates everything over an event broker (a system that routes messages between components). It comes with built-in support for sessions, artifacts, authentication, evaluators, and observability.

Current state of the repository

The actual source code is not here yet — the README says it is "coming to this repository soon." Right now, the repo holds two things:

  • Agent Mesh Desktop — a free application you install on your computer
  • Ready-to-run examples — sample use cases you can apply

Agent Mesh Desktop

A single application that runs on your machine and includes a command-line tool called sam. Supported platforms:

  • macOS 11 (Big Sur) and later
  • Windows 10 and later (x86_64)
  • Linux (marked as Experimental)

Installation is done via a script (curl on macOS/Linux, PowerShell on Windows) that downloads the app from the latest release. The script checks the download's SHA-256 checksum before installing, and verifies Apple notarization on macOS or the Authenticode signature on Windows. You can preview what the script would do with a --dry-run flag instead of actually installing.

The sam CLI

After installation, you use sam from a terminal. Common commands:

  • sam ai-assistance skill install — installs skills that teach AI coding assistants (like Claude Code or Codex CLI) how to build agents, tools, and configuration for your Agent Mesh version. Skills go to .agents/skills/ by default; use --target claude for Claude Code.
  • sam config plan -m manifest.yaml — previews a configuration change before applying it
  • sam config apply -m manifest.yaml — applies a configuration change to a running Agent Mesh instance
  • sam docs — opens the documentation offline

The ai-assistance and config commands are marked Early Access, so their behavior may change between releases.

Examples

The examples directory contains use cases you apply with sam config apply. One described example: a support ticket triage scenario where a customer creates a ticket in a fictional "Acme Help Desk," and Agent Mesh classifies the issue, calculates the response deadline, drafts the first reply, escalates urgent tickets, and writes the result back to the ticket. The README notes these examples are for learning and evaluation, not production-ready.

Production

Agent Mesh Desktop is only for building and testing on your own computer. Production use requires a commercial license from Solace. For production, you deploy Agent Mesh to Kubernetes, or use Agent Mesh Cloud (a managed cloud service).

Other notes

  • License: The repository uses the BUSL-1.1 license (from the SPDX header), with per-part licensing details in a NOTICE file.
  • Contributions: The repo does not accept issues or pull requests. Questions go to the Solace Community; bugs and security issues go to support@solace.com.
  • First-time LLM connection: To connect a large language model the first time you open the app, see the linked docs page ("Installing the Desktop Bundle").

All facts above are drawn directly from the README; nothing has been added or inferred beyond what it states.

Related

  • Project
    Martian-Engineering/lossless-claw

    A plugin for the OpenClaw framework that prevents context loss by using a DAG-based summarization system to store and retrieve conversation history in a local SQLite database.

  • Project
    TheoLeeCJ/SemIf-OpenJev

    An open-source framework for making fast, typed semantic decisions by reading model logits directly instead of generating text, reducing latency and parsing overhead for AI agents.

  • Project
    chaitin/MonkeyCode

    MonkeyCode is an open-source AI office workbench (product name: 百智办公 MonkeyAI) where teams describe tasks in natural language and AI delivers finished office documents, code, or designs. It includes team resource sharing, usage billing, permission auditing, and usage statistics. It can be self-hosted with a one-command install (min 1 CPU + 1 GB RAM), supports bringing your own models and MCP-connected tools, and offers desktop clients for Windows, macOS, and Linux. Free tier covers up to 10 seats; paid tier adds cloud PC and support. Licensed under AGPL-3.0.

  • Project
    EvoScientist/EvoScientist

    EvoScientist is a multi-agent research framework that automates the scientific lifecycle by using self-evolving memory, autonomous skill generation, and a team of specialized sub-agents to perform complex research tasks.