Technology Sharing | How to quickly build OpenAI-compatible Agent services using the Agno framework

Today I will share with you a technical demo, showing how to quickly build an OpenAI-compatible API service based on the Agno framework , and implement functional expansion through MCP tool calls . The greatest value of this solution is that the client does not need to be modified in any way , as long as it is compatible with the OpenAI API, it can directly call the configured tool capabilities!

https://github.com/lemonhall/agno_mcp_openai_demo

Cherry studio configuration

The effect of operation

Functional description of mcp service

Note: caldav-mcp has not been released yet due to functional testing.

Core Values ​​?

In traditional Agent development, the client needs to:

1. Understand tool calling conventions
2. Handling complex tool registration processes
3. Implement tool calling and result analysis

The solution demonstrated by this project: ✅ The client only needs to call the standard OpenAI API ✅ All tool call logic is encapsulated on the server ✅ No need for the client to mount any MCP tools ✅ Maintain full API compatibility

Technical implementation analysis?

1. Architecture Design

The project adopts a three-tier architecture:

1. API compatibility layer : fully simulate the OpenAI Chat Completions interface
2. Agent core layer : Agent implementation based on Agno framework
3. Tool service layer : integrating MCP tool services (taking CalDAV as an example)

Client → OpenAI compatible API → Agno Agent → MCP tool service

2. Key technical points

• OpenAI API fully compatible : support synchronous/streaming responses
• Transparent tool call : the client does not need to be aware of the existence of the tool
• Flexible tool configuration : unified management tool registration on the server side

3. Code Structure

.
├── simple_openai_agent_api.py   # Main service entry
└── agno_openai_adapter.py       # OpenAI adapter layer implementation

Quick experience ?

1. Install Dependencies

pip install fastapi uvicorn agno mcp

1. Start the service

python simple_openai_agent_api.py

1. Calling example (using any OpenAI compatible client)

response = client.chat.completions.create(
    model = "agno-agent" ,
    messages=[{ "role" :  "user" ,  "content" :  "Check my schedule" }]
)

Scene expansion?

Although the demo uses calendar management as an example (because this scenario is simple and easy to understand), this solution is applicable to:

• Database query tools
• E-commerce order management
• Internal system integration
• Any scenario that requires tool calling

Design concept ?

1. Separation of concerns : The client only cares about the business dialogue, not the tool implementation
2. Smooth migration : Existing OpenAI applications can be migrated seamlessly
3. Flexible expansion : The server can dynamically increase or decrease tool capabilities

Conclusion

This demo demonstrates the unique advantages of the Agno framework in building enterprise-level Agent services. Through the OpenAI compatibility layer, we can:

1. Greatly reduce client access costs
2. Maintain API stability and compatibility
3. Flexible expansion of backend tool capabilities