evaluations
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides instructions for the user to install companion components (
experimentsandonline-evaluations) from the vendor's repository using thenpx skillsutility. - [INDIRECT_PROMPT_INJECTION]: The skill classifies user intent to route requests to other specialized skills, creating a potential surface for indirect prompt injection.
- Ingestion points: User-provided requests for evaluations processed in SKILL.md.
- Boundary markers: The instructions include logic to detect ambiguity and mandate a clarification question to the user before proceeding.
- Capability inventory: The routing skill itself lacks dangerous capabilities like file writing, network operations, or command execution; it functions solely as a hand-off mechanism.
- Sanitization: Uses intent classification to restrict routing to two predefined paths.
Audit Metadata