Plan mode
Questions first, then an editable plan, then Build. No surprises in a large change.
Most coding agents stop when the code is written. Nexus keeps going: it runs the tests, opens your app in a real browser, clicks through the flow it changed, and hands you screenshots, a recording and a code review.

Every UI change gets a browser check with a screenshot per step and a recording you can watch. If a check fails, Nexus fixes the cause and runs it again before it hands off.

Questions first, then an editable plan, then Build. No surprises in a large change.
A second agent reviews the diff for bugs, security issues and missing tests.
Try several approaches on separate branches and keep the best one.
Risky commands wait for approval, and every change is a diff you can keep or undo.
Follows AGENTS.md and rule files, and remembers decisions across sessions.
It pauses instead of failing when credits run out or the server restarts, then continues from the same step.
It works inside a full cloud machine with your running app, so it can run tests and check the UI in a real browser, and it shows that evidence with every change.
Yes. Choose Manual, Auto or Full auto. In Manual and Auto, risky actions such as destructive commands or database writes wait for your approval. Full auto runs them without asking (blocked commands stay blocked).
No. Secrets are encrypted and injected into your app; the agent only works with their names, and values are removed from anything it reads.
Yes. Background agents run in the cloud and open a pull request when they finish. You can also schedule agents.
Open a project in your browser and let Nexus help you build it. Your first month is free.
$0 today · Cancel any time · Your code stays yours