think
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- EXTERNAL_DOWNLOADS (LOW): The skill requires installing
google-generativeaivia pip. This is a trusted dependency from thegoogleorganization; thus, per the [TRUST-SCOPE-RULE], the finding is downgraded to LOW.\n- PROMPT_INJECTION (LOW): Surface for indirect prompt injection detected (Category 8).\n - Ingestion points: User input enters the agent context through placeholders such as
[Your problem],[Your question], and[The idea to challenge]in theSKILL.mdtemplate.\n - Boundary markers: Absent; there are no delimiters or explicit instructions provided to the LLM to ignore potentially malicious commands embedded within the user-provided text.\n
- Capability inventory: The skill demonstrates capability to perform network operations through the
geminiCLI tool which contacts external APIs.\n - Sanitization: Absent; the skill lacks escaping or validation logic for the external content interpolated into the prompt templates.
Audit Metadata