nebula-visual-verification

What problem does it solve?

Provides a guarded, deterministic workflow for focused visual verification of changed Canvas components and pages so reviewers can reliably detect contrast, overflow, and obvious image/copy regressions without running exploratory full-app QA.

Core Features & Use Cases

• Deterministic signal collection: Requires running a browser-side helper to capture contrast ratios, overflow metrics, and image context as machine-readable evidence.
• Scoped review: Targets specific Workbench preview routes, authored mock states, and a tight set of viewports (desktop, tablet, mobile) to avoid noisy, unfocused checks.
• Guided remediation loop: Includes an auto-fix and reverify loop, a checklist-driven pass/fail rubric, and explicit stuck/closeout rules for clear outcomes.
• Use Case: A frontend engineer submits a component change and requests visual verification; the reviewer starts Workbench, runs the helper for each named state and viewport, and produces a concise deterministic closeout with artifacts only when needed.

Quick Start

Start or reuse Canvas Workbench, open the changed preview route and named state, and run the collect-visual-signals.js helper via agent-browser for each desktop, tablet, and mobile viewport before declaring results.