Your first scenario

Write a test as one English sentence, let the agent verify it in a real browser, and watch Scout record a deterministic script you can replay forever. You state the outcome you care about; Scout proves the product clears that bar — and re-checks itself when the UI moves.

This walkthrough assumes you’ve installed Scout and scout doctor reports ✓.

1. Describe the flow

A scenario is a heading and a plain-language sentence. Write what a user does and what must be true — never selectors or waits. Create .scout/specs/search.scout.md:

---
feature: Search
---

## Homepage search returns results
On the homepage, search for "shoes" and a list of results appears.

That’s the whole test. No fixtures, no data-testid, nothing to keep in sync with the markup — the agent discovers the real elements at run time. (Prefer not to hand-write it? Your coding agent can author a richer spec straight from repo context — see AI agents & MCP.)

2. Run it

scout go

On this first run an agent drives a real browser, confirms the outcome holds, and records exactly what it did as a deterministic script. The recording is born from actions that actually worked — not from a guess about your DOM.

3. Read the verdict

scout report

Every run resolves to a single verdict — verified, partial, failed, or blocked — each mapped to a stable exit code. verified means the flow held and a script is now cached. See Verdicts & exit codes for the full contract.

4. Replay is free

Run scout go again. This time there is no agent and no API call — Scout replays the recorded script in milliseconds. You pay for judgment once, on the first run; every run after is pure, fast, deterministic replay.

When the UI later drifts and a step breaks, Scout re-aims itself: the agent returns, re-verifies your intent, re-records the path, and hands you a partial to sign off. No human babysitting the suite. That unattended heal is what makes Scout safe to point at code an agent just wrote.

---

Next: make it a CI gate with Environments & CI, or go deeper on the format in Writing scenarios.