Agents

AMI never tests your app alone. It deploys agents with one job each, and each agent writes every action to the run log so you can see who did what.

• Requirements Analyst turns requirements into testable acceptance criteria.
• Test Designer plans cases across every suite in your config.
• UI Runner drives the running app in a real browser.
• API Prober checks endpoints, contracts and error paths.
• Regression Guard re-runs what changed and everything it touches.
• Data Checker verifies writes, migrations and data integrity.
• Reporter writes the verdict with repro steps and severity.

Agents share one plan but work independently, so a slow browser test never blocks an API check. AMI waits for every agent before it reports a verdict.

Each agent's actions are recorded with a timestamp and the target it touched, for example agent-03 passed checkout/flow. See Run Log.