Skip to main content

Reliability Model

How Agent Mantis achieves reliable test execution at scale.

Flakiness reduction​

Agent Mantis reduces test flakiness through:

  • Auto-waiting — Never interacts with elements before they're ready
  • Self-healing selectors — Adapts when UI changes
  • Smart retries — Distinguishes transient failures from real bugs
  • AI interpretation — Understands intent, not just DOM structure

Failure analysis​

When a test fails, Agent Mantis provides:

  • Screenshot of the page state at failure
  • Step-by-step execution log
  • Suggested fixes (when possible)

Monitoring​

Track test reliability over time in the Portal dashboard:

  • Pass rate trends
  • Flaky test detection
  • Average run duration