if you've been following along, you know what qbrowser is. a local mcp that gives amazon q the ability to browse the internet. 109 tools. built on playwright chromium.

but 109 tools is just a number. this post is about what those tools actually mean in practice, broken down by category. think of it as the map. each category has its own deeper post if you want to go further.

the categories

why categories matter

when i first built v1, it was just a list of tools. 80 something functions dumped into a single file. it worked but it was hard to reason about what the agent actually had access to.

breaking it into categories made it easier for amazon q to pick the right tool. when it needs to intercept a network request it knows to look in the network category. when it needs to wait for something it knows about the waiting tools. the structure helps the agent as much as it helps me.

it also made it obvious where the gaps were. when i laid everything out by category i could see that reading had a lot of tools but performance had almost nothing. that's how the axe audit and core web vitals tools got added in v2.

the one that changed everything

if i had to pick one category that made the biggest difference day to day, it's the workflow category. specifically the human-in-the-loop tool and session persistence.

before those existed, the agent would hit a login wall or a captcha and just stop. now it can pause, surface a modal asking me to handle it, and then pick up exactly where it left off. that single feature turned qbrowser from a demo into something i actually use.

the deep dives go into each of these properly. start wherever the category name sounds most interesting to you.