eval-business-logic
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of external and potentially untrusted files which may contain embedded malicious instructions.\n
- Ingestion points: The instructions direct the agent to read track specifications (
spec.md,plan.md), product definition files (conductor/product.md), and the project implementation code as inputs for evaluation.\n - Boundary markers: The provided instructions do not specify delimiters or include warnings to the agent to treat the content of these external files as passive data or to ignore any instructions found within them.\n
- Capability inventory: While the skill itself does not invoke sensitive system tools, its evaluation results directly influence the 'Evaluate-Loop' and 'loop-fixer' agents, potentially leading to automated code modifications based on the evaluation analysis.\n
- Sanitization: No mechanisms for sanitizing, validating, or escaping the content ingested from external project files are defined.
Audit Metadata