Proteus Workflow Engine

What is Proteus Workflow Engine?

Proteus Workflow Engine is a Python-based workflow engine built on FastAPI that supports multi-agent systems, automated workflows, and MCP server integration. It provides intelligent agents, over 20 built-in node types, and a web visualization interface. Designed for developers building complex, automated task pipelines.

How to use Proteus Workflow Engine?

Clone the repository, create a Python 3.11+ virtual environment, install dependencies from requirements.txt, copy .env.example to .env, and set your LLM API key. Start the service locally with python proteus/main.py or using Docker (docker-compose -f proteus/docker/docker-compose.yml up -d). Access the web interface at http://localhost:8000. Workflows can be defined in YAML config files.

Key features of Proteus Workflow Engine

• Over 20 built-in node types including MCP client, handoff, browser agent
• Intelligent Agent system with Chain-of-Thought reasoning
• Real-time execution monitoring via Server-Sent Events (SSE)
• Web visualization interface with multiple interaction modes
• Docker support for containerized deployment
• MCP (Model Context Protocol) support for external tools and resources
• Sandbox environment for secure code execution

Use cases of Proteus Workflow Engine

• Automated research workflows: search, crawl, analyze, and export reports
• Multi-agent collaboration for complex task decomposition
• Weather data retrieval via MCP-connected map services
• Browser automation for web scraping or form filling
• Custom workflow nesting and looping for process automation

FAQ from Proteus Workflow Engine

What platforms and LLM models does Proteus Workflow Engine support?

It runs on Python 3.11+ and can be deployed locally or via Docker. The LLM is configurable; the default model is Deepseek Chat, and you can set MODEL_NAME and optional REASONER_MODEL_NAME in the .env file.

Does Proteus Workflow Engine support MCP servers?

Yes. It includes an MCP manager that can dynamically load remote tools via SSE. You can configure MCP servers in proteus_mcp_config.json and use the mcp_client node type or the “MCP智能体” agent mode.

What is the licensing for Proteus Workflow Engine?

The project is licensed under the MIT License, as stated in the included LICENSE file.

What are the main known limitations or planned improvements?

Planned enhancements include workflow pause/resume, version control, additional node types (e.g., email, file format conversion), multi-agent memory, and cluster deployment support. These are listed in the README’s TODO section.

Can I create custom node types?

Yes. You can add new node types by inheriting BaseNode in src/nodes/ and registering them in node_config.yaml. The README provides a code example for doing so.