fix(visualizer): improve dark mode styling for ShinyText component

[quanru]

Summary

This PR improves the dark mode visual experience for the ShinyText component and related UI elements in the visualizer package.

Changes

ShinyText Component (Dark Mode)

• Replace solid purple text color (#c5b8ff) with a vibrant animated gradient:
  • Colors: #a78bfa → #c084fc → #e879f9 → #c084fc
  • Maintains the smooth 8-second gradient animation
• Replace harsh white shine effect with subtle purple shine:
  • Color: rgba(197, 184, 255, 0.15) (15% opacity)
  • Much softer and more harmonious with the text gradient
  • Preserves the left-to-right sliding animation

PlaygroundResult Component (Dark Mode)

• Add dark mode styling for loading progress text
• Add dark mode styling for pre code blocks with proper background and border

Type Safety Improvements

• Add null/undefined checks in replay-scripts.ts
• Add type normalization for ExecutionDump handling
• Improve error handling with console warnings

Visual Comparison

Before: Solid purple text (#c5b8ff) with white shine (40% opacity) - visually jarring in dark mode

After: Vibrant purple-pink gradient with matching purple shine (15% opacity) - harmonious and elegant

Test Plan

• Run npm run lint - passed with auto-fixes
• Build packages successfully
• Verify dark mode styling in browser DevTools
• Test animation smoothness

Related

Addresses user feedback about harsh white shine effect in dark mode while maintaining the dynamic visual appeal of the component.

[netlify]

❌ Deploy Preview for midscene failed. Why did it fail? →

Name	Link
🔨 Latest commit	7b6570c
🔍 Latest deploy log	https://app.netlify.com/projects/midscene/deploys/693fd73ec24e19000824b5c4

[quanru]

Additional Fix

Added a second commit to fix a bug in resolveAiActionContext that was caught by unit tests:

Issue: The function was incorrectly prioritizing preference.aiActionContext over target.aiActionContext, which contradicted the documented behavior.

Fix: Corrected the resolution order to:

1. target.aiActionContext (highest priority - per-target override)
2. preference.aiActionContext (fallback - agent-level preference)
3. preference.aiActContext (fallback - deprecated alias)

Test: All Puppeteer unit tests now pass (12/12), including the previously failing ai-action-context.test.ts.

[quanru]

Update: aiActionContext Priority Changed

I've updated the aiActionContext resolution logic based on feedback:

New Priority Order (Highest to Lowest)

1. preference.aiActionContext - Agent-level preference (highest priority)
2. preference.aiActContext - Deprecated alias (fallback)
3. target.aiActionContext - Target-level context (lowest priority)

Changes Made

• ✅ Updated function comments to reflect correct priority
• ✅ Modified test cases to validate new behavior
• ✅ Added comprehensive test coverage (5 test cases)
• ✅ All tests passing

Rationale

This design allows agent-level preferences to override target-level context, providing more flexibility for runtime configuration while maintaining backward compatibility with the deprecated aiActContext alias.

[quanru]

Correction: aiActionContext Priority

I've corrected the priority order. preference.aiActionContext is the deprecated property, not aiActContext.

Correct Priority Order (Highest to Lowest)

1. preference.aiActContext - Agent-level preference (recommended, highest priority)
2. preference.aiActionContext - Agent-level preference (deprecated)
3. target.aiActionContext - Target-level context (lowest priority)

Changes Made

• ✅ Updated function comments to correctly identify deprecated property
• ✅ Modified test cases with accurate naming
• ✅ All tests passing (5 test cases)

Rationale

This design:

• Prioritizes the new recommended aiActContext property
• Maintains backward compatibility with deprecated aiActionContext
• Allows agent-level preferences to override target-level context
• Provides flexibility in runtime configuration