overview
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOW
Full Analysis
- SAFE (SAFE): No malicious patterns, obfuscation, or data exfiltration techniques were found. The skill consists entirely of instructional markdown documentation.
- Architectural Safety (INFO): The skill promotes safety best practices by defining a 'Courtroom' mental model where the agent is treated as 'untrusted' and must record all reasoning and evidence for its actions.
- Indirect Prompt Injection Surface (LOW): The skill describes an interface for reading external data products (e.g., customers_v1). Ingestion points: External data read via the 'decision_read' tool. Boundary markers: None specified in this overview. Capability inventory: Includes 'decision_read', 'decision_write', and 'decision_evaluate'. Sanitization: Not addressed in this foundational guide. While the described capabilities create a potential surface for indirect injection, the framework itself is designed as a mitigation layer to provide auditability for such risks.
Audit Metadata