Autonomous Agent Scoping — Define Safe AI Boundaries

I will assess your autonomous agent requirement and deliver a scoping report defining the appropriate autonomy level, safe action boundaries, risk classification for each action type, oversight requirements, and a phased autonomy expansion approach starting with maximum human oversight and progressively expanding agent autonomy as reliability is demonstrated. Autonomous agents operating without clearly defined action boundaries consistently exceed their intended scope — the agent's LLM reasoning will identify actions that are within its tool permissions but outside the intended design scope, and without explicit boundary definition, it will take them.

The report covers autonomy level assessment (L1-L5 scale), action boundary definition, risk classification per action type, oversight checkpoint design, and a phased autonomy expansion plan with reliability gates.