Question: Any plans to make the specification more OpenAI compatible ?

### Is your feature request related to a problem? Please describe.

Unnecessary message conversations because of different schemas.

### Describe the solution you'd like

I know, it's probably weird to ask, but I saw that Google have an OpenAI compatible specification, can we use this for the messages structure instead of the parts API ?

It's probably a huge breaking change, but better early then never :)

Frankly when I look on the schema of most providers they have the messages following one spec.
The spec of Google is just different with message parts - IMO it creates unnecessary overhead and computation when I want to integrate the project not just with Gemini LLMs.

To give you an example:

I might want to submit a task to an A2A server that has an LLM, that LLM doesn't necessarily have to be a Gemini one, then if I want to interact with other vendor's API's I have to convert the messages before sending them to their API (most of them use OpenAI compatible spec).

### Describe alternatives you've considered

I think it could be way simpler if we just use the same spec or maybe in a future ADK / SDK introduce a message conversation as a util that people could use for language processing with other LLMs. WDYT ?

### Additional context

Goal is, having a vendor agnostic and broader compatibility.

### Code of Conduct

- [x] I agree to follow this project's Code of Conduct

Look at the current message structure in the A2A protocol specification and compare it with the OpenAI API schema. Understand the breaking changes involved in migrating from the parts API to a messages-based structure. Review existing integrations and how they handle message conversion.

Escrito por el modelo de indexación a partir del texto del issue.