v0.2.0 — M7: LLM Integration Frontend ↔ Backend

- api.js: API bridge to FastAPI backend
- Dynamic NPC dialogues via gemma4:12b (cached 30s)
- Periodic world event generation (every ~10s)
- Event validation (no soft-lock)
- Backend health check on startup
- Fallback to static dialogue if LLM offline
- Tested: events ✅, dialogues ✅

CHANGELOG.md

@@ -28,6 +28,26 @@
 - NPC dialogue is static (LLM not yet wired to frontend)
 - Only one map (Eichenhafen)
 
-### Next Milestone
-- M2: Refine pathfinding, add map transitions
-- M3: Wire LLM dialogue to frontend NPC interactions
+## v0.2.0 — M7: LLM Integration Frontend ↔ Backend (2026-07-17)
+
+### Added
+- API bridge (`api.js`) connecting frontend to FastAPI backend
+- Backend health check on startup — shows LLM model name in event log
+- Dynamic NPC dialogues via Ollama (gemma4:12b)
+  - Each NPC gets unique dialogue based on personality, mood, and player class/level
+  - 30-second dialogue cache to reduce LLM calls
+  - Fallback to static dialogue if LLM unavailable
+- Periodic world event generation (every ~10 seconds)
+  - LLM generates optional events (weather, discoveries, encounters)
+  - Events validated by backend (no soft-lock possible)
+  - Events displayed in event log with ✨ prefix and purple color
+- NPC interaction now shows "LLM generating..." placeholder while waiting
+
+### Changed
+- NPC interaction is now async (waits for LLM response)
+- Event log color coding: purple (llm), red (combat), green (quest), yellow (system)
+
+### Tested
+- Backend health endpoint: ✅
+- Event generation: ✅ (gemma4:12b generates atmospheric events)
+- NPC dialogue: ✅ (personalized, in-character responses)