PydanticAI A2A integration

[sergioave]

Question

I see in the documentation that communication between pydanticAI agents via the a2a protocol is done through FastaA2A, but seeing that there have been no changes for two months and no responses to issues, is this still the method recommended by pydanticAI, or should we migrate to a2a-sdk for communication between remote agents?

Additional Context

No response

[Mahdi-CV]

I have the same question here

[CostcoFanboy]

IMO for FastA2A to be MVP, A2A protocol compliance would need to be pushed to 0.3.
pydantic/fasta2a#28

Given how stretched Kludex's time is, despite how admittedly cluttered a2a-sdk is, I think the FastA2A should leverage the official a2a-sdk.

[Kludex]

I'll make FastA2A up-to-date.

Given how stretched Kludex's time is, despite how admittedly cluttered a2a-sdk is, I think the FastA2A should leverage the official a2a-sdk.

I think I tried it before. It didn't work as I would expect...

---

Does someone have a good UI to test an A2A server? One that doesn't require me to spin up container e.g. https://github.com/a2a-community/a2a-ui .

[Kludex]

This one also forces me to spin up containers: https://github.com/a2aproject/a2a-inspector