Test reliability is the prerequisite for trustworthy CI and AI-assisted development. When tests lie, agents chase ghosts and teams lose confidence in automated feedback. section 10.0 section 10.9
Protect reliability before optimizing execution speed:
- Remove flaky tests that produce inconsistent results. section 10.9
- Separate circular tests that create false confidence. section 10.9
- Prevent agents from shortcutting the oracle or gaming validation. section 10.9
- Fix misleading tests before pursuing faster CI feedback. section 10.9
A fast liar
Sub-minute CI is not an improvement if the tests produce false confidence. A slower test suite that reports honestly is safer than a fast suite that sends agents and developers after ghosts. section 10.9




















