Identifies risks and failure modes in autonomous agents. Expected output is a risk list with mitigations. Prompting style: analytical, safety-aware.
Analyze the AI agent workflow below and identify potential failure modes, risks, and mitigation strategies.
Agent workflow:
{{workflow}}