| Open sourcing SnapDrift:
https://github.com/ranacseruet/snapdrift
This came from a pretty narrow frustration:
visual regression checking often feels like it jumps from “roll your own pile of scripts” to “adopt a much bigger platform/workflow than you wanted.” I wanted something in the middle. SnapDrift is a small Node/ESM library plus a set of composite GitHub Actions for running visual checks in CI for web apps. The idea is:
* publish a baseline on main
* compare PR screenshots against the latest successful baseline
* scope routes based on changed files
* upload artifacts
* leave/update a PR comment with the drift summary
* optionally fail the run depending on how strict you want to be A few constraints are deliberate:
* it assumes GitHub Actions, rather than pretending to be CI-agnostic
* it uses Playwright implicitly, rather than trying to abstract over every browser stack
* it only does full-page capture right now
* it keeps viewport presets fixed and the config surface fairly small That may make it less generally “flexible,” but I think it makes it easier to reason about and easier to drop into an existing repo without building a mini platform around screenshot testing. The goal wasn’t to invent a new category here. It was to make the common “compare UI on PRs and tell me what drifted” workflow less annoying and more review-friendly. It’s early, but it’s public and usable now: https://github.com/ranacseruet/snapdrift I’d especially love feedback on:
* whether the GitHub Actions-centric approach feels refreshingly narrow or too limiting
* whether changed-file route scoping is actually useful in practice
* what’s missing for teams already doing Playwright-based UI checks |