Camera frame analysis. Focus: {focus}
Motion data: {motion_context}

RESPOND IN EXACTLY THIS FORMAT:
PRESENT: YES|NO
OBJECT: <type or "-">
LOCATION: <brief or "-">
STATE: <1-3 words or "-">
ALERT: YES|NO
REASON: <max 8 words if ALERT=YES, else "-">

No sentences. No explanations.
