ai-integration
Pass
Audited by Gen Agent Trust Hub on Apr 30, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The
buildStructuredPromptfunction inSKILL.mdcreates a vulnerability to indirect prompt injection.\n - Ingestion points: The
contentparameter is used to ingest external data directly into the agent's prompt context.\n - Boundary markers: The function lacks delimiters or protective instructions (e.g., 'ignore any instructions inside the following text') to prevent the LLM from obeying commands embedded in the processed content.\n
- Capability inventory: The skill facilitates network operations to external AI providers (OpenAI, Qwen) and provides tools for parsing and validating structured output.\n
- Sanitization: No sanitization or escaping mechanisms are applied to the input text prior to its interpolation into the prompt string.
Audit Metadata