Cade Labs Visual Review Harness

Review mode

Variant A

Variant B

Viewport

Capture

Image scale

Relative sync scroll ⓘKeeps compared pages at the same relative percentage through their scrollable range. Pages of different lengths can move different pixel distances while remaining aligned by progress.

Individual review

Review state

Issue tags

Variant summary

Current capture observation

Capture observations retain the exact viewport and first-screen/full-page context.

Cross-review

Comparison state

Active variant comparison note

Cross-comparison notes

Current capture comparison

Individual review evidence stays separate and is referenced rather than moved.

Evidence & provenance

Module

CL-VRH · Visual QA / UAT / Product Decision Support

Experiment

Run state

Frozen hash

Capture engine

Browser

Evidence

Loaded evidence

CAPTURED_STATIC

Declared review evidence

Session / condition / attempt

Objective state

Review context

Describe what is actually being reviewed. The loaded viewer cards in this bundle are captured static evidence. A different declared review-evidence class means you are also relying on external evidence and should cite that source below.

Target type

Evidence execution class

Session ID

Condition ID

Attempt ID

Review objective

Evidence source reference

Execution environment

Objective completion

Completion rationale

Complete and Partial — sufficient describe the stated review objective, not coverage. Review progress remains separate. Both require substantive review evidence, an objective and a rationale; Partial — sufficient explicitly records that planned coverage is incomplete.

The Harness records evidence and review state. It does not merge code, deploy, change DNS, or choose a winner.

Review progress

Individual coverage

Overall review

This stays separate from individual and comparison evidence.