test-explore
Installation
SKILL.md
Exploration
A program drives the interface at random and, after every action, checks properties: rules that must hold whatever the user did. It reports the shortest action sequence that breaks one, and the report replays. Scripted end-to-end tests prove the paths someone wrote down; exploration checks the rules on the paths nobody did.
Choose a driver
| Driver | Fits | What you get |
|---|---|---|
| Bombadil | Web, Electron, terminal programs; any stack | Temporal properties (always, eventually.within, next), recorded traces, the bundled runner and page kit |
| fast-check + Playwright | JavaScript or TypeScript projects already on Playwright | Model-based sequences inside the existing runner, shrinking to minimal reproductions, seed-and-path replay |
| Hypothesis + Playwright | Python projects | RuleBasedStateMachine rules and invariants, shrinking, a failure database |