The app you work in: conversation that edits code, a board that dispatches to your agents, and the settings that actually apply.
Screen by screen
Code
A conversation that edits your repository, with the real history of what it did — every tool call, every file touched, and the option to revert.
Editor
A real editor with tabs and a file tree: syntax highlighting, the open file in the URL so a link can point at it, and a refusal to edit a file that was only partly read.
Work
Runs and receipts: per attempt, what approved it and what it actually changed; the diff left in your repository; and, across a project's runs, what each configuration cost and whether it was worth it.
Knowledge
Memory, profile and skills — what the agent has kept, what it concluded about you from that, what it learned to do, and what is waiting for your review before it is used.
Automation
Schedules, a task board in columns from backlog to done — each card naming the agent that picks it up — the registry of those agents, and projects that run against a spec until the spec says they are done.
Settings
Models, memory, autonomy, tools and your agent's own name, language and standing instructions. Most take effect on the next call; the few that wait for your next conversation, or for the next time you start the app, say so on the row.
Install it
Native installers for Windows, macOS and Linux. No Python needed — the backend is bundled.