Objective
Add a reference workflow that demonstrates visual regression testing using playwright + cache-memory for baseline screenshot storage. This addresses the lowest-scoring scenario (FE-1, 3.8/5.0) from the Agent Persona Exploration discussion #18250.
Context
The agent correctly suggests playwright for screenshots but leaves baseline comparison storage unaddressed. Without a reference workflow showing how to store "before" screenshots using cache-memory or artifacts, visual regression workflows are incomplete.
Approach
- Create
.github/workflows/visual-regression.md as a reference workflow that:
- Triggers on
pull_request
- Uses
playwright for capturing screenshots
- Uses
cache-memory to persist baseline screenshots between runs
- Compares current screenshots against stored baselines
- Uses
safe-outputs to post a comment with the diff report
- Follow the pattern of existing reference workflows like
auto-triage-issues.md
- Ensure filesystem-safe timestamp formats (no colons) in cache-memory filenames
Files to Create/Modify
- Create:
.github/workflows/visual-regression.md (and recompile to generate .lock.yml)
Acceptance Criteria
Generated by Plan Command for issue #discussion #18250