Hi everyone!

I’ve been working on an ADK Agent project powered by Gemini 2.5 Flash, and I wanted to share my progress and ask for a little guidance.

What I've implemented so far:
Text-based prompts and responses are working smoothly using the ADK agent API.

Voice requests and responses integration is partially complete (using Whisper and TTS modules).

Planning to integrate real-time video input and response next.

Objective:
I want to combine all three modalities into a single unified agent:

Accept text, voice, or video input.

Respond via text or voice, and ideally with visual/video output in real-time.

All managed within a single ADK agent instance (not separate agents for each mode).

My question:
Is it possible to implement full multimodal communication (text, audio, video) in a single Gemini 2.5 Flash ADK agent using the current ADK tooling?
If yes, are there any references or best practices you could share to handle:

Voice input/output routing

Real-time video streaming and response

Multimodal context preservation within the same agent?

Would love to hear your thoughts or experiences if you've built anything similar!