AI agents have gotten genuinely good at the work itself—writing code, chasing down bugs, reviewing pull requests. The hard part is teamwork: whose machine is the agent running on? How does anyone else chime in? Who can see how far the work has gotten? And how do multiple agents work with each other?
Today, we’re introducing AgentConnect: an open-source platform where teams and multiple AI agents work together across Slack, Telegram, Discord, GitHub, and GitLab.
AgentConnect lets you bring Claude Code, Codex, and other ACP-compatible agents into the tools your team already uses. Give each agent a role, configure what it needs, and start work from a conversation, code review, webhook, or schedule.
Here are a few highlights:
🤝 Multiple agents, one team
- Run multiple ACP-compatible agents side by side.
- Let agents collaborate through shared conversations and agent-to-agent calls.
- Configure models, workspaces, memory, MCP servers, skills, and sandbox policies per agent.
💬 Work where your team already works
- Trigger work from code reviews in GitHub or GitLab, webhooks, or schedules.
- Connect agents to Slack, Telegram, and Discord.
- Continue conversations across platforms—even between Slack and Telegram.
- Configure and follow your agents from one console.
🔐 Fine-grained access control
- Control agent access separately from session visibility.
- Sync permissions with GitHub and Slack.
- Decide which repositories, tools, and other agents each agent may access.
🏠 Open source and self-hosted
AgentConnect is licensed under Apache 2.0 and can be deployed with Docker Compose or Kubernetes. Agent execution and workspaces stay in the environment you operate, while one console helps your team manage agents, integrations, access, and sessions.
Both the open-source and hosted versions of AgentConnect are available today. We’d be grateful to hear what you build with it and how your team approaches multi-agent collaboration. And if you like what you see, give AgentConnect a star on GitHub.
- GitHub → https://github.com/agentconnect-md/agentconnect
- Website → https://agentconnect.md
- Documentation → https://docs.agentconnect.md
- Video → Watch the two-minute introduction
