What's New:

• Smart block messages — When a message is blocked, the overlay now explains WHY the data is sensitive and WHAT to do
  instead. Per-category guidance for PII, credentials, API keys, health data, financial data, secrets, and internal terms.
• Warning hints — Warning banners now include a fix suggestion so you know how to rephrase safely.
• LLM topic classification — If your admin has defined sensitive topics (e.g. "customer financial data"), prompts are now
  classified against those topics using AI in addition to pattern matching.
• AI Tool Policy enforcement — If your organization has enabled the AI Tool Restriction Policy, the extension respects the
  approved/blocked tools list sent from the dashboard.

Improvements:

• Redesigned block overlay with per-violation detail cards
• Scrollable violation list for messages with multiple detections
• Matched text shown in monospace for clarity
• Cleaner visual styling on block and warning UI

v1.0.13 Changes:
- No new permissions required
- Bug fix release — no functional changes to the extension itself
- Server-side DLP scan improvements: smart pattern detection now correctly activates all rules when enabled by org admin
- Fixed CVV detection pattern to handle natural language phrasing
- Fixed international phone number detection regex
- Extension version bump for consistency with app release