synapse-ai

Build AI agents that actually do things. 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, trading stocks, and anything else you can wrap in an MCP server or n8n workflow.

---

git clone https://github.com/naveenraj-17/synapse
cd synapse
docker-compose up

Open http://localhost:3000

npm install -g synapse-ai
synapse

Python 3.11+ required. All Python dependencies are installed automatically on first run.

pip install synapse-ai
synapse

# or with uv:
uvx synapse-ai

Node.js 18+ required.

---

Most AI agent frameworks give you a loop and a few toy tools. Synapse gives you a production-grade platform:

• ReAct reasoning engine — agents think, act, observe, and iterate up to 30 turns per task
• 8 built-in tool servers ready to use out of the box
• Plug in any MCP server — local stdio or remote HTTP, added in seconds
• Build custom tools with n8n — turn any automation workflow into an agent tool with zero code
• Orchestrate multiple agents as a DAG — parallel branches, routing logic, loops, human checkpoints
• Persistent vault — agents save and share files across sessions and runs
• Local-first — runs entirely on your machine with Ollama, or connect any cloud LLM

---

Synapse agents are powerful because of what they can do. Every tool is a separate MCP process — isolated, composable, and safe.

These run automatically when Synapse starts:

| Tool Server | What It Does | |---|---| | Sandbox | Execute Python code in an isolated Docker container (512 MB RAM, 1 CPU). Pre-loaded with pandas, numpy, matplotlib, scikit-learn,