Generate a Maestro test
Describe what the test should prove and rnx writes the flow:
The CLI inspects the app currently open in rnx, generates a short Maestro flow from your goal (taps, inputs, waits, and a final assertion that proves the goal), runs it immediately, uploads the recording as a preview, and registers the run in your dashboard. The output prints the result, the run id, and the preview link.
Two things make this different from an agent poking at your app:
- +The output is deterministic. You pay for authoring once; what runs is a plain flow with real assertions. Keep it in
.maestro/and it replays forever without another model call, exactly like a hand-written flow. - +The result is hosted. The run’s recording, per-step trace, and pass/fail land where your team already looks, alongside every other run.
Requirements
Maestro generation calls a model through rnx cloud, so it needs a
signed-in CLI (rnx login) or a RNX_API_KEY with the llm scope,
and it draws from your plan’s included usage. The preview upload rides the
same run.
Attribute runs to your account
Pass the repo so the run indexes under your account’s dashboard rather than your personal account: