[Feature Request] All worker could accept multi-modal information

[Wendong-Fan]

Motivation

now only multi-modal agent has the toolkit to read multi-modal information, but now many LLMs support vision, it's better to natively support worker agent get image path from decomposed sub tasks

Solution

No response

Alternatives

No response

Additional context

No response

[lightaime]

Adding to this: the workers should take multi-modal information as input messages.

[bittoby]

@Wendong-Fan @Pakchoioioi Can I pick this issue up, too? I can handle this. thank you

[Wendong-Fan]

thanks @bittoby ! issue assigned to you

[Wendong-Fan]

hey @bittoby , just a quick catch up, any updates or anything we can support to move this issue forward?

[bittoby]

@Wendong-Fan I am doing that, too. I will send pr soon