historical-source-adapter
Pass
Audited by Gen Agent Trust Hub on May 14, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection (Category 8) by interpolating raw user input into the agent's instructions.
- Ingestion points: User-provided text enters the context through the
original_source,student_level, andtarget_skillsplaceholders inSKILL.md. - Boundary markers: There are no structural delimiters or unique markers surrounding the user-provided placeholders to isolate them from the agent's core logic.
- Capability inventory: The skill's capabilities are restricted to text generation and formatting; it does not request or use tools for filesystem access, network connectivity, or shell execution.
- Sanitization: The input is not sanitized or validated for instruction-like patterns before being processed by the model.
Audit Metadata