qa-sweep

Automate code review for feature implementation and bug fixes by orchestrating parallel sub-agent implementation followed by independent adversarial review. This structured workflow enforces cross-checks and a 95% confidence gate to surface regressions before finalizing work.

An adversarial review workflow for quality assurance is a process where independent sub-agents cross-check feature implementation to surface edge cases. It imposes explicit phases including parallel implementation and completeness verification to achieve high confidence in delivery.

Run a project-wide QA sweep by initiating parallel sub-agent implementation, then triggering independent adversarial review agents. The process applies mandatory guardrails and a completeness verification phase to validate task completion with at least 95% confidence.

Yes, the 95% confidence gate stops task completion for debugging errors if independent adversarial review fails. It enforces mandatory review phases and guardrails to ensure high-quality outcomes before allowing any debugging or failure investigation to finalize.

The best way to trace tech debt and investigate failures is using parallel sub-agent implementation combined with independent adversarial review. This approach applies a structured completeness verification phase to ensure high confidence before stopping the investigation.