Report Agent Risk Data

Populate data_risk metadata on span types for an agent that is already instrumented with Prefactor.

Core principle: infer, don't guess. Read the code each span type wraps and reason about what data enters and leaves it — then record that as risk metadata.

Trigger Phrases

Apply this skill when the user asks for any of these:

• "add risk data to my agent"
• "populate data_risk for span types"
• "instrument risk data for compliance"
• "what data does my agent handle?"
• "fill in risk metadata on spans"
• "configure data governance for my agent"

Prerequisite