synapseorchai/synapse-ai repository overview

⁠Synapse AI

Synapse AI is a privacy-first, open-source multi-agent orchestrator designed for creating, connecting, and managing AI agents. It empowers AI models to execute complex, real-world tasks through an extensible Model Context Protocol (MCP) tool ecosystem.

Whether you are running local models or connecting to cloud providers, Synapse allows agents to seamlessly chain actions—like web browsing, code execution, and database querying—to build robust, automated pipelines.

⁠Custom Tooling: The MCP Powerhouse

Synapse AI turns your existing infrastructure into agent-ready tools in minutes. You no longer need to write complex integration logic for every new capability.

• API to MCP: Instantly convert any REST API or Webhook into a functional MCP tool. Simply point the orchestrator to your endpoint, and your agents can start interacting with your external services.
• Python to MCP: Turn any legacy or custom Python script into an agent tool. If you have a script that processes data, generates reports, or interacts with local hardware, Synapse wraps it in a secure MCP layer for agent execution.
• Zero-Boilerplate Integration: Focus on the logic of your tools while Synapse handles the protocol, execution environment, and state management.

⁠Enterprise & Internal AI Solutions

Synapse AI is designed to be the "brain" sitting on top of your company's internal data and tools.

• Privacy-First Architecture: Deploy Synapse entirely on-premise or within your VPC. Keep sensitive company data behind your firewall by utilizing local models (via Ollama/vLLM) and locally-hosted tool servers.
• Unify Siloed Data: Use agents to bridge the gap between Jira, GitHub, Slack, and internal SQL databases. Build agents that can "investigate a bug report in Jira, check the relevant code in GitHub, and summarize the fix for the team in Slack."
• Internal Process Automation: Automate complex internal workflows like employee onboarding, automated code reviews, or real-time financial data synthesis without sending data to third-party providers.
• Safe Execution: Use the built-in Docker Sandbox to allow agents to write and run code against internal datasets without risking the host system's security.

⁠Key Features

• Native Orchestration Builder: Instantly generate the underlying Directed Acyclic Graph (DAG) for multi-agent workflows just by chatting with the AI Builder.
• Enterprise-Ready Docker Integration: Spin up isolated environments in minutes. Designed for businesses to seamlessly plug into internal AI pipelines without environment conflicts.
• Local-First & LLM Agnostic: Run entirely locally using Ollama or connect to cloud providers (OpenAI, Claude, Gemini).
• ReAct Reasoning Engine: Agents utilize a "think, act, observe, and iterate" loop to handle complex, multi-turn tasks efficiently.
• Advanced DAG Routing: Build workflows with parallel branches, conditional routing logic, loops, and human-in-the-loop approval checkpoints.

⁠Tool Ecosystem & MCP Integrations

Synapse treats every tool as an isolated, secure Model Context Protocol (MCP) process. Convert any Python script or webhook into an agent tool instantly without writing integration boilerplate.

Included Native Tools:

• Python Sandbox: Isolated Docker container for executing code with pre-loaded data science libraries (pandas, numpy, etc.).
• Web Scraper & Browser: Full browser automation (Playwright/crawl4ai) capable of bypassing anti-bot measures.
• SQL Agent: Introspect schemas and run read queries on PostgreSQL, MySQL, and SQLite.
• File Vault & Parsers: Persistent file sharing across sessions with built-in PDF and Excel parsing.
• Code Search: Semantic semantic search across indexed repositories.

⁠Tech Stack

• Backend: Python (3.11+)
• Frontend/Server: Node.js (22+)
• Tool Management: uv / uvx for MCP server execution
• Environments: Docker (Sandbox & Enterprise Deployment), Chromium (Browser automation)

⁠Getting Started with Docker

Deploying Synapse AI is easier than ever with our official Docker support:

# Pull the latest image
docker pull synapseorchai/synapse-ai:latest

# Run the container
docker run -d -p 3000:3000 synapseorchai/synapse-ai:latest