E2E tests can be generated automatically now. Argent lets you record Flows (sequences of interactions) and replay them later. Generating tests becomes part of the loop, making sure UI regressions never happen. 🛡️ Try it out now with the /argent-create-flow skill!

Ahah watching agent device and argent tweeting the same feature the same day since day one is kind of funny 😁 Amazing tools both

By by maestro :P

Wow

We LOVE Argent at @AwesomeLabsLLC

How quick is the replay

Recording flows to auto generate tests is a smart workflow.

Does this support maestro?

Recording flows instead of writing tests manually is a 10x shift. Mobile interaction patterns are too complex for manual E2E.

hu, nice one. but we fully automated via , it takes fully lifecycle

Recording flows kills the writing cost, which is the part everyone complains about. The maintenance cost is the harder half, since a recorded flow pins itself to whatever selectors existed at record time and the first redesign turns the whole library red at once. Does replay resolve elements by role or accessible name, or by the DOM path it captured?

Can we collectively create a standard spec for writing these flows that are agnostic to the runner? I vote no proprietary files, let’s make a standard maestro, argent, agent-device, X tool’s runner all perform well at. Determinism is necessary.

is this the same as maestro mcp or dif?

This is cool