Rill Browser - The browser where Claude Code and Codex work beside you
Rill is an AI-native browser built around the Claude Code and Codex you already use. Browse normally, then press ⌘E on any page to turn what you're looking at into a task for your agent. Rill finds the right project, passes along the context, and the agent works while you keep browsing. It also sorts your tabs into themes, turns your history into days you can search, and brings every project your agents are working on into one place. Free, on the plan you already have.
Replies
Haha my reader later tab just got nervous
@jojo_li hahah I have so many of those read-it-later tabs too. That's the goal - closing them confidently by turning them into action
Interesting! Is it a built-in AI that forwards the tasks to Claude and Codex or do you directly integrate your already existing Claude or Codex subscriptions API keys into Rill? Congrats on the launch! 🚀
You know when you’re reading something and suddenly think I could build this? This feels pretty useful for that.
@kyle_bennett6 Right? There are so many times I felt that, reading papers, scrolling through X.com, watching videos etc. This is largely why I built Rill, so that moment of "suddenly think I could build this" doesn't go away but becomes concrete improvement. Imagine how that'll multiply
Turning the page you are already looking at into a task for your agent with one shortcut is a really natural way to blend browsing and building. Passing along the right project context is the part that usually goes wrong in these setups, so handling that automatically is nice. Does the agent stay in the background while you keep browsing other tabs?
sending an idea straight from the page makes sense. does the task keep a snapshot of that page, so the agent still has the original context if i navigate away or the site changes?
Congrats on the launch, William! The code-switching problem is real. Ideas tend to hit when I'm reading something unrelated, and having ⌘E turn the page into a task for the agent in the right project sounds like a great way to catch them. Looking forward to trying it out.
the per-turn snapshot and Undo you mentioned in another thread is the detail that got me, I've been asking a version of this question on a few launches lately - does that snapshot/undo only cover the git state, or does it also know about anything the agent did outside the repo during that turn (an API call it made, a form it submitted on the page it was reading)? code is easy to revert, a request that already fired isn't, and that gap is the part I haven't seen anyone solve yet