Anvil Loop Specialist Workflow

Overview

Anvil Loop formalizes the rigorous, multi-stage process of ensuring a feature is measurably complete and matches its initial specification (the 'Gold Star'). It moves beyond subjective sign-offs by enforcing a structured loop where multiple specialist agents repeatedly test against fixed criteria until all automated tests pass and visual discrepancies are resolved.

Capabilities

• Structured Quality Gates: Enforces a multi-step completion checklist, ensuring no stage is skipped (Spec $\rightarrow$ Strike $\rightarrow$ Measure $\rightarrow$ Record $\rightarrow$ Reflect).
• Automated Measurement: Integrates objective testing by running XCUITest suites and performing pixel-by-pixel screenshot diffing against golden references.
• Continuous Iteration: Manages the loop, automatically identifying failure points (failed test rows or visual deviations) and directing subsequent agent runs to address those specific gaps.
• Knowledge Capture: Automatically logs all failures into a LESSONS.md file for continuous process improvement.

Example Use Cases

• Complex UI Implementation: When building a new dashboard component, Anvil Loop ensures that every interaction (e.g., tapping, scrolling, data loading) passes both unit tests and visual fidelity checks against the design mockups.
• Integration Testing: After connecting two major system modules, it can be used to verify that the handoff points function correctly under various simulated loads until all edge cases are covered.
• Feature Parity Check: If a feature needs to behave identically across iOS and Web views, this agent ensures both platforms meet the exact same measurable standard.