🚀 Feature Requests - Vote with 👍

[vvincent1234]

Share your feature ideas and vote with 👍 on what you want most!
We prioritize development based on community feedback and upvotes.

[thegcode1337]

An improvement that would be really helpful is being able to interrupt, aka provide feedback to the agent while it is working live. Example when you see it struggle to find the right place to click or whatnot, just being able to type directly to give it extra details/guidance when needed.

I did notice that you can pause then terminate then enter new commands then restart and it does seem to remember to history, it just didn't feel intuitive and I actually surprised that that work (thinking "terminate" would be like canceling the job) , so I guess technically we already can do what I'm asking it would just be to make it a bit more clear and intuitive UI wise, like pause, enter some input, then resume.

Thanks!

[vvincent1234]

An improvement that would be really helpful is being able to interrupt, aka provide feedback to the agent while it is working live. Example when you see it struggle to find the right place to click or whatnot, just being able to type directly to give it extra details/guidance when needed.

I did notice that you can pause then terminate then enter new commands then restart and it does seem to remember to history, it just didn't feel intuitive and I actually surprised that that work (thinking "terminate" would be like canceling the job) , so I guess technically we already can do what I'm asking it would just be to make it a bit more clear and intuitive UI wise, like pause, enter some input, then resume.

Thanks!

I just released a version where I changed the name from "cancel" to "pause" to avoid confusion. Yes, I understand what you mean — currently, you can pause and then resume, but the feature doesn’t yet support inputting new extra details or guidance. I plan to add this support in the next version.

[vvincent1234]

An improvement that would be really helpful is being able to interrupt, aka provide feedback to the agent while it is working live. Example when you see it struggle to find the right place to click or whatnot, just being able to type directly to give it extra details/guidance when needed.

I did notice that you can pause then terminate then enter new commands then restart and it does seem to remember to history, it just didn't feel intuitive and I actually surprised that that work (thinking "terminate" would be like canceling the job) , so I guess technically we already can do what I'm asking it would just be to make it a bit more clear and intuitive UI wise, like pause, enter some input, then resume.

Thanks!

Supported now, please try the new version.

[Dexterward75]

Hello team. Would it be possible to shorten the display of agent profiles? When we put a few lines of profile per agent, it quickly becomes cumbersome in the display of our agents ?. Thank you.

[vvincent1234]

Hello team. Would it be possible to shorten the display of agent profiles? When we put a few lines of profile per agent, it quickly becomes cumbersome in the display of our agents ?. Thank you.

Could you show me the screenshot of this case?

[Dexterward75]

Hi. I think I confused the "profile description" with the classic profile of an agent where you have to inform them about their role/context/task... And as a result I had descriptions that took up the entire window for a single agent in the "LLM" category. I realize this is actually just a brief description of the agent and not his personal prompt. Sorry for the inconvenience .

[ain3sh]

@vvincent1234 Firefox compatibility anytime soon? <3 Would love to use with Zen browser

[vvincent1234]

@vvincent1234 Firefox compatibility anytime soon? <3 Would love to use with Zen browser

Hi, thank you for your attention. VibeSurf is built on Chrome DevTools Protocol (CDP) instead of Playwright. But Firefox has discarded CDP since one certain version, so we cant support Firefox now.