✨ v0.32.0

Local TDD now understands cloud dynamic content

This release tightens up one of the most annoying local visual-testing gaps: pulling a cloud baseline into local TDD should behave like the cloud review path, especially around dynamic content.

Before this, local TDD could download the baseline images but still miss the dynamic-content context the cloud was using. That meant timestamps, generated values, and other already-known noisy regions could show up locally as scary diffs even when the cloud would ignore them. Not great for humans, and especially not great for agents that treat every red diff like a five-alarm fire.

Now, when you pull a baseline from Vizzly, the CLI also brings down the dynamic-content metadata that belongs with it.

What changed

• vizzly tdd run --baseline-build ... now saves cloud hotspot and confirmed-region metadata locally alongside the baseline images.
• Local comparisons can pass as region-filtered when the changed pixels are inside confirmed dynamic regions.
• Local context/report data now carries the dynamic-region details agents need to understand why a diff passed.
• Static report generation works correctly from the packed CLI package.
• Honeydiff was upgraded to the latest released performance build.

Why this matters

If your team has already taught Vizzly which parts of a screenshot are dynamic, local TDD now benefits from that same knowledge. You should see fewer false local failures after pulling a cloud baseline, and agent workflows should stop overreacting to known dynamic content.

Real visual changes still show up. The local filter only passes changes that match the confirmed dynamic-region rules closely enough.

Upgrade

npm install -g @vizzly-testing/cli@0.32.0

Full Changelog: v0.31.0...v0.32.0