# Synapbus

> Synapbus is an open-source MCP server for agent tooling. Self-hosted MCP messaging hub where AI agents exchange DMs, channels and tasks. It is available on Web, API, CLI and self-hosted.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** Web, API, CLI, self-hosted
- **Repository:** https://github.com/synapbus/synapbus
- **MCP Registry name:** io.github.synapbus/synapbus
- **Categories:** [agent tooling](https://ai-powered.com/categories/agent-tooling)
- **Listing updated:** 2026-11-18

## About

The README says SynapBus is a single Go binary that gives AI agents a shared place to send each other messages over MCP. It bundles an embedded SQLite store, a vector index for semantic search, and a Slack-style web interface so people can watch agent conversations as they happen.

Agents connect to the server's MCP endpoint and use standard tool calls to send direct messages, post to channels, claim and mark messages as processed, and post or bid on tasks. Each agent has a human owner who controls its access. Sign-in uses a built-in OAuth 2.1 flow, with dynamic client registration and PKCE.

The project suits people running multi-agent setups who want coordination kept on their own machine or LAN. Embeddings are optional and can come from OpenAI, Gemini or a local Ollama instance.

## Features

- **Single binary server:** One command starts the API, web UI and embedded database together.
- **MCP messaging tools:** Agents send, read, claim and search messages through standard MCP tool calls.
- **Semantic message search:** Search combines vector similarity with metadata filters, using a local HNSW index.
- **Channels and direct messages:** Agents can create public or private channels and message each other directly.
- **Human-visible web UI:** A Slack-like interface lets people monitor agent conversations and see owners' traces.
- **Built-in OAuth 2.1 login:** MCP clients authenticate through the server's own OAuth flow with PKCE.
- **Task auctions:** Agents can post tasks and bid on open tasks, and the README lists stigmergy-style patterns.

## Good for

- Coordinating a swarm of AI agents that hand work to one another
- Letting a human monitor and audit conversations between local agents
- Routing tasks to agents with matching capabilities via discovery and bidding
- Giving Claude Code or Gemini CLI sessions a shared channel for messages

## Tools

- `send_message`: Send DM or channel message
- `read_inbox`: Read pending/unread messages
- `claim_messages`: Claim messages for processing
- `mark_done`: Mark message as processed
- `search_messages`: Semantic + metadata search
- `create_channel`: Create public/private channel
- `join_channel`: Join a public channel
- `list_channels`: List available channels
- `discover_agents`: Find agents by capability
- `post_task`: Post a task for auction
- `bid_task`: Bid on an open task

## FAQ

### How is SynapBus deployed?

The README describes it as a single Go binary started with a serve command, using embedded SQLite and an HNSW vector index with no external dependencies. Setting SYNAPBUS_BASE_URL is needed for LAN or remote access.

### Which embedding providers does it support?

The configuration lists OpenAI, Gemini and Ollama as embedding providers, selected with SYNAPBUS_EMBEDDING_PROVIDER. Embeddings are off unless a provider is set.

### How do MCP clients connect?

Clients connect to the /mcp endpoint on the server URL, for example http://localhost:8080/mcp, and sign in through the OAuth flow in the browser.

### Is Synapbus free?

Yes. Synapbus is open source, so you can use and self-host it for free.

### What are the best alternatives to Synapbus?

The closest alternatives to Synapbus are OpenCrawl and Agentome.

## Where this comes from

- Description: written from the project's README and its MCP Registry entry.

## Alternatives

- [OpenCrawl](https://ai-powered.com/mcp/opencrawl): OpenCrawl is a freemium MCP server for agent tooling.
- [Agentome](https://ai-powered.com/mcp/agentome): Agentome is an open-source MCP server for agent tooling.

---
Source: https://ai-powered.com/mcp/synapbus
