aibdd-red-evaluate

To validate a failing BDD test report, you evaluate the runner-native evidence against feature files and step definitions to distinguish a legal, meaningful red failure from a false-red environment issue or a hollow-red non-behavior gap. This prevents downstream execution from consuming invalid signals.

A false red in BDD workflow orchestration occurs when test failures stem from runner errors, invalid step definitions, or environment issues rather than a genuine user-observable behavior gap. It is detected by enforcing hard gates on runner-native test report evidence before proceeding.

You stop Green execution by applying an evaluation gate to the Red handoff payload, which reads test reports and step definitions to emit a FAIL verdict with veto-level findings if false-red or hollow-red conditions are detected. This blocks the RED → GREEN flow.

You need to evaluate step definitions before proceeding to Green when a Red Worker run completes, ensuring the failing evidence is legally grounded in valid scenario-to-step mappings and establishes a real behavior gap. This evaluation outputs a PASS/FAIL report.

Yes, you can use test report analysis to detect hollow-red non-behavior failures by applying semantic detection gates that flag assertions against non-public internals or harness issues. This ensures the red failure establishes a real user-observable behavior gap.

Automated red evaluation is limited to reading only allowed evidence boundaries within runner-native test reports and runtime-visible files, meaning it cannot assess external state or ungrounded scenarios. It strictly enforces Type A and Type B hard gates based solely on visible artifacts.