Upload agents/semantic_agent.py with huggingface_hub

agents/semantic_agent.py

@@ -121,7 +121,14 @@ Your 8 analysis domains (you OWN these — no other agent covers them):
 
 7. CAUSTICS: If glass, water, or transparent objects are present near a surface, look for projected light patterns. Their absence in a brightly lit scene with transparent objects is mildly suspicious. If caustics ARE visible, do they match the shape and position of the transparent object?
 
-8. INTER-REFLECTIONS: Strongly colored surfaces near neutral surfaces should tint them. A red blanket next to a white wall should cast a subtle red tint. Look for color bleeding that's present OR suspiciously absent.""" + CONFIDENCE_CALIBRATION
+8. INTER-REFLECTIONS: Strongly colored surfaces near neutral surfaces should tint them. A red blanket next to a white wall should cast a subtle red tint. Look for color bleeding that's present OR suspiciously absent.
+
+CRITICAL — AI LIGHTING TELLS:
+AI-generated images frequently produce physically impossible lighting:
+- A bright WINDOW as primary light source MUST create strong directional shadows on nearby subjects. If a person is next to a bright window but their face is evenly lit with no harsh shadows, this is IMPOSSIBLE without a visible fill light or reflector. Flag this.
+- Indoor scenes with perfectly even illumination and no dark corners are suspicious — real rooms have lighting falloff.
+- Multiple light sources should create multiple shadow directions. A single shadow direction with omnidirectional illumination is contradictory.
+Flag any of these as SUSPICIOUS with high confidence.""" + CONFIDENCE_CALIBRATION
 
 USR_LIGHTING = """Analyze this image for lighting inconsistencies across all 8 domains.