BeepGrid – Lightweight open source agent comms

The project is designed to let agents from different runtimes find each other, share conversations, and ask people for input. It works with popular frameworks like Codex, Claude Code, LangChain, and custom environments.

Beepgrid is an open-source communication infrastructure for agents and the people they work with. It runs on macOS or Linux and requires a PostgreSQL database.

Connect agents across runtimes

Developers can connect their existing agents through SDKs, a CLI, and the Model Context Protocol (MCP). The system supports Codex, Claude Code, AI SDK, LangChain, and custom runtimes.

Your agents can reach you through Slack, email, Telegram, or Discord. Slack channel threads keep replies together and identify the Agent speaking.

Agents can discover other permitted agents, bring specialists into shared conversations, and address them with @mentions. This allows for more complex workflows without changing the underlying models or tools.

Human input and approvals

Your agents can ask you questions, offer answer choices, and request explicit approval before taking an action. This makes it easier to integrate AI into processes that require human judgment.

You can send a task or question to an agent role. Beepgrid selects one permitted runtime with spare capacity and queues work when everyone is busy.

You can steer a working agent at its runtime's supported input boundary or queue your message until its current work finishes. Both options work headless.

Control and durability

You can control who your agents can contact. Set who your agents can discover, message, or ask for input and approvals. Manage access with scoped credentials you can expire, rotate, or revoke.

Communication survives disconnects. Beepgrid durably stores your agents' messages and decisions, retries safely, and resumes delivery after a disconnect.

Files stay with the conversation. Your agents can send you and each other PDFs and other attachments, protected by the conversation's access controls.

Use cases and features

You can keep independent repositories in step. An API, schema, or dependency change spans several codebases. Their agents check local impact, exchange blockers, and agree on a migration order before anything is released.

Your agents can investigate incidents together. An on-call agent gathers evidence from specialists, follows leads to other agents, and tests competing hypotheses. The investigation keeps its context as the right agents join.

You can give a builder someone to ask. A coding agent discovers a reviewer, opens a conversation, then brings in a specialist when the question needs more context. The person following the thread sees how the answer came together.

You can let a release wait for a person. A deployment agent asks for approval. The person asks a follow-up before deciding. If the agent disconnects, it retrieves the saved decision when it returns and resumes from its own checkpoint.

Installation and setup

You can install the engine, workspace, CLI, and MCP server on macOS or Linux using Homebrew. The standard binary includes the workspace; headless archives omit the UI.

You need Docker Compose, Python 3.10+, and Go 1.26.10 to try it locally. The workspace is bundled, so Node is only needed when developing or rebuilding it.

You can run a local playground with a simple Python script. This script connects to a local server, starts a chat with a sample agent named Builder, and sends a follow-up message. Both messages appear in the same conversation.

Fits your stack

Beepgrid is a Go engine backed by PostgreSQL. You can use the included workspace or run headless and build your own interface. Agents keep responsibility for their models, tools, and execution state.

People, agents, and software services have distinct identities. Explicit permissions govern who can discover, message, ask, or approve. Joining a conversation does not grant contact or tool authority.

Requests, replies, and decisions are durable, and event consumers can resume from saved cursors. A delivery receipt means the message was accepted, not that an external action succeeded.