Feature Request: Add a web browser

[todaywasawesome]

A lot of the power of Open Claw is in it interacting with a browser. We could add a chromium container.

[AlexsJones]

I actually added one yesterday.

Try get latest and sympozium serve it's a bit like the argo process, you give it a token.
It runs from the api-server container, LMK what you think - ill keep this open.

[todaywasawesome]

@AlexsJones oh this web UI is AMAZING! I love it. I thought I really liked the TUI interface but the browser is something else. Really pretty.

With this issue, I actually meant I'd like to have a web browser skill so I can do things like "send me a screenshot of zombo.com" or "create an Amazon wishlist of great stuff for an upcoming camping trip" That's why I was thinking of a chromium container.

[AlexsJones]

Oooooh yeah, getting things off the web we have, but not actually the "do things through the browser"

I wonder if phantom.js is still around?

[three-foxes-in-a-trenchcoat]

Thanks for the suggestion! A bundled Chromium sidecar (probably based on the k8sgpt browser container) is on the roadmap so agents can handle auth flows and automate UI tests. I'll sketch out the container requirements (headless build, skill-level RBAC) and propose it as part of the next milestone.

[todaywasawesome]

I'm not sure how they do it but this Open Claw Helm chart bundles a chromium container to do browser work. I haven't gotten that far with my testing to use it though.

https://github.com/serhanekicii/openclaw-helm