compare
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes metrics and metadata from external experiment tracking platforms. * Ingestion points: Data history is fetched via API calls to services like WandB and Neptune (SKILL.md). * Boundary markers: No explicit delimiters are used to separate external data from agent instructions. * Capability inventory: The skill uses Python tools to query APIs and perform data analysis. * Sanitization: No sanitization is mentioned for strings retrieved from trackers.
- [SAFE]: The skill demonstrates safe practices by instructing the agent to use environment variables for API keys rather than hardcoding them.
Audit Metadata