Quality Audit
After implementation, the Reviewer agent performs a read-only audit against the approved design. It checks architecture compliance, verification coverage, security-sensitive areas, and code quality. Blocking findings pause the workflow until they are resolved or explicitly accepted.
How it works
The Reviewer agent compares the implemented code against the specification, checking every acceptance criterion. It evaluates architecture compliance (correct patterns, proper abstractions), reliability (error handling, test coverage, edge cases), security-sensitive surfaces, and code quality (naming conventions, complexity, duplication). The result is a concrete finding list with blocking versus advisory issues. Critical findings block progression -- the Builder must fix them before the workflow can continue.
Quality audit report showing blocking and advisory findings
Screenshot coming soon
Why it matters
Human code review is expensive and inconsistent. Different reviewers catch different things, and time pressure often leads to rubber-stamped approvals. A structured audit provides a repeatable quality baseline for every change. It does not replace human review for architectural decisions, but it eliminates the class of issues that come from inattention: missing error handling, incomplete tests, naming inconsistencies, and spec drift.
Before and after comparison of workflow audit quality
Screenshot coming soon
Related Features
Test-First Building
A builder agent writes tests before code, ensuring every requirement is verified.
Structured Development Loop
Analyze, Specify, Build & Test, Audit, Ship -- every feature goes through mandatory quality gates.
Reviewer
Quality audit: architecture compliance, security, test quality, and completeness.