configuring-experiment-analytics
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing untrusted user input to modify external experiment configurations.
- Ingestion points: Natural language user requests for experiment metrics and event names ingested in SKILL.md.
- Boundary markers: The instructions do not explicitly require the use of delimiters or 'ignore' warnings when the agent interpolates user input into tool parameters.
- Capability inventory: The skill utilizes sensitive capabilities including 'experiment-update' (modifying experiment settings) and 'experiment-saved-metrics-create' (creating shared project metrics).
- Sanitization: The skill mitigates risks by requiring the agent to call 'read-data-schema' to verify event existence and 'experiment-saved-metrics-list' to check for existing definitions before proceeding with updates.
Audit Metadata