OpenWebUI-OpenAI-Agent is a multi-agent pipeline implementation based on the OpenAI Agent SDK. It is designed to provide flexible multi-agent conversation and tool-calling capabilities for OpenWebUI. The project supports multiple models and agent collaboration, suitable for general Q&A, complex reasoning, and code generation scenarios.

• Multi-agent collaboration: Built-in General Agent, Reasoning Agent, and Main Agent for automatic task routing.
• Tool calling: Supports custom tool (FunctionTool) registration and asynchronous execution.
• Streaming output: Real-time event streaming for better user experience.
• Extensible: Easily integrate custom models, tools, and pipeline logic.

• openai_agent_sdk_pipe.py: Main pipeline implementation, including agent definitions, tool registration, and event stream handling.

• General Agent: Handles general questions and basic coding tasks.
• Reasoning Agent: Handles complex reasoning and advanced coding tasks.
• Main Agent: Automatically routes user requests to the appropriate agent.

MIT License.

This project is built with the OpenAI Agent SDK. For SDK usage and advanced features, please refer to the official documentation.