What’s missing from rhun for your AI coding workflow?
Hey everyone, following up on our launch with a question about where rhun should go next.
The idea behind rhun is a small, fast editor for working alongside coding agents. You can already run agents in the terminal, follow Claude Code and Codex sessions in the agents panel, and review changes with built-in Git tools.
I'm especially interested in two things:
AI integrations: Which tools do you use, and what would you want rhun to do with them? Send selected code to an agent, manage sessions, make reviewing agent changes easier, or something else?
Missing features: If you’ve tried rhun, what made you switch back to your usual editor? What’s the one feature you couldn’t comfortably work without?
Concrete examples would help us decide what to build next. Tell us what you were trying to do, which tools were involved, and where you got stuck.
Thanks for giving rhun a try!


Replies
What I'd actually use is a diff since the agent's last turn, not since the last commit. I don't commit between turns, so by round three git diff is one blob and I can't tell what the agent just did from what it did an hour ago.
Additionally, list the files each session has touched right in the agents panel. That's what I check before letting a second agent start
@siarheihamanovich hi Siarhei! That's a great food for thoughts. My plan was adding a new "Lifecycle" panel to incorporate this functionality (along with others additions). Still thinking about the UX. Thanks for the input, I'll keep you posted about the progress.
Nice input! A diff since the last agent turn would save a lot of review load, I’ve had a broad git diff hide the change that broke the last workflow. Seeing touched files beside the session would make second passes much safer.
@matheus_paranhos1 heya Matheus. Thanks for trying rhun! It looks like the diff overview is something we need to implement next (see my comment above).
I haven’t tried rhun yet. Building Turnfeed with Codex, avatar cropping and likes that preserve scroll position needed browser checks as well as diff review. I’d find it useful to keep the intended visible behavior and the observed check result beside a turn’s changes. Could the Lifecycle panel accommodate that—for example, a screenshot and a note that scroll position was actually checked?