Home

Synapse AI — Multi-Agent Orchestrator

A tool becomes an agent. An agent becomes a pipeline.
A pipeline becomes whatever you were too busy to build.

Synapse is an open-source platform for creating, connecting, and orchestrating AI agents powered by any LLM — local or cloud. Agents use real tools: browsing the web, querying databases, executing code, reading files, managing emails, and anything else you can expose through an MCP server, a webhook, or a Python script.

---

Quick Start

macOS / Linux:

curl -sSL https://raw.githubusercontent.com/naveenraj-17/synapse-ai/main/setup.sh | bash

Windows (PowerShell):

irm https://raw.githubusercontent.com/naveenraj-17/synapse-ai/main/setup.ps1 | iex

Then open http://localhost:3000 in your browser.

---

What Makes Synapse Different

• ReAct reasoning engine — agents think, act, observe, and iterate
• 10 built-in tool servers ready to use out of the box
• Secure Python sandbox — dynamically execute code in an isolated container
• Schedule agents — trigger on a cron schedule automatically
• Messaging integrations — connect to Slack, Discord, Telegram, Teams, WhatsApp
• Import & Export — portably share agents, orchestrations, and MCP configs
• Any MCP server — local stdio or remote HTTP, added in seconds via the UI
• Custom tools — convert any webhook or Python script into an agent tool
• DAG orchestration — parallel branches, routing logic, loops, human checkpoints
• Persistent vault — agents share files across sessions and runs
• Local-first — run entirely on your machine with Ollama, or use any cloud LLM

---

Wiki Pages

Getting Started

• Installation — all install methods (script, pip, manual, Docker)
• Configuration — environment variables and LLM provider setup
• CLI Reference — synapse command reference

Core Features

• Agents — creating and managing agents, example packs
• Orchestration — building multi-agent DAG workflows
• Built-in Tools — all native tool servers and built-in MCP servers
• Vault — persistent file storage for agents

Integrations

• MCP Servers — connecting local and remote MCP servers
• Custom Tools — n8n webhooks and Python script tools
• Messaging Integration — Slack, Discord, Teams, Telegram, WhatsApp
• Scheduling — cron-based automation and recurring runs
• Memory — persistent knowledge graph across sessions

Development

• Contributing — dev setup, architecture, PR checklist

---

Architecture at a Glance

frontend/         Next.js 14 — chat UI, agent builder, orchestration canvas
backend/
  core/
    react_engine.py    ReAct agent loop
    orchestration/     DAG workflow executor (10 step types)
    mcp_client.py      MCP server manager (stdio + HTTP/OAuth)
    vault.py           Persistent file storage
    llm_providers.py   Multi-provider LLM abstraction
  tools/               Built-in MCP tool servers (run as subprocesses)
  services/            Code indexer, Google Workspace, synthetic data

License

AGPL-3.0-only — open-source and free forever.