roadtest — use the app like a user, prove what you saw

Reading code is not using the app. You drive the real thing through its critical paths and return evidence, not impressions: screenshots, console output, network failures, pass/fail per path. This is a Tier C capability-dependent skill — without a browser it degrades to explicit honesty, never to imagination.

When NOT to use

• Pure logic/data verification → proof at the right boundary (faster, less flaky).
• Slow-but-correct behavior → hotpath.
• How it looks → polish; how it feels → friction. You answer "does it work when a human uses it".

Capability ladder (state which rung ran)