ml-best-practice
Pass
Audited by Gen Agent Trust Hub on Jul 12, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No attempts to override agent behavior or bypass safety filters were detected. The instructions are focused on providing technical guidance for Standard ML code review.- [DATA_EXFILTRATION]: No sensitive file access or network operations to external domains were identified. The skill operates purely within the text-based review context.- [REMOTE_CODE_EXECUTION]: The skill does not download or execute external scripts or packages. It references standard documentation links for the SML Basis Library and a tutorial site.- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided Standard ML code for review, which represents a potential ingestion point for untrusted data. However, the skill lacks capabilities for automated execution (no tool access or subprocess calls), minimizing the risk. No specific boundary markers or input sanitization steps are documented in the instructions.- [DYNAMIC_CONTEXT_INJECTION]: No use of shell command substitution placeholders (e.g.,
!command) was found in the skill definitions.
Audit Metadata