Scripting - creating cross page macros, like you now can do with puppeteer etc. Have simple basic programming capabilities. Stuff like that now shows up with “AI” agentic browsers, but that’s too. much. I just want to set up macros, like “go to my timesheet page, click start, enter current time”. On a “Autohotkey for the web” level. (Power users instead of developers)

Tab management - I’m working a lot with jira and other “wonderful” software. What would be nice would be showing multiple tabs at once (like opening several browser windows) , also maybe automatically creating a conceptual “tree” in the tab overview. That would require some configuration (on top of what normal vertical tabs do). For example confluence has a implicit tree, why shouldn’t the tab overview in the browser track that. A lot of web sites are ordered hierarchically. The only tab hierarchichy we currently have in the browser is a “i opened tab b from tab a” hierarchy

History search - not using it is proof it doesn’t fulfill a desire. “Damn I recently was on a site that talked about how the confabulator works without causing wobbles in the swingmode arm” Trying to find that after you did open a few hundred other pages sucks. It would be nice to have positive and negative search terms, have a “near” operator etc. So that would be a full text search engine (which already exists) about pages I have seen in the past.

Granular permissions: I only allow a page to enumerate the fonts it needs to use, not all of them so it can calculate a hash. I want to forbid it from accessing certain domains (Adblockers can currently do that) etc. You may/may not play video. The permission system is in place, but too coarse.

And yes there are privacy containers, but not in a really helpful manner yet. They’d need to integrate with the above permissions, for example so I can put a web page into a jail of its own.

All these aren’t well thought out features, rather I pulled them out of my butt. Still I feel there’s close to no innovation on the usability of a browser and we could need that. We still have the same interaction model as in the 90s with Mosaic - and while (of course) not every idea would work out in a good way, some things would be worth following up on. I’d expect that out of an organization like Mozilla (less so out of commercial browser vendors).