regex-vs-llm-structured-text
Pass
Audited by Gen Agent Trust Hub on Apr 3, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The
validate_with_llmfunction inSKILL.mdcontains an indirect prompt injection surface where untrusted document content is directly embedded into the validation prompt. This could allow an attacker to place instructions within a document that override the parsing logic. - Ingestion points: The
original_textparameter invalidate_with_llm(derived from the sourcecontentof the document being processed). - Boundary markers: Absent; the prompt uses simple string interpolation (f-strings) without delimiters like XML tags or clear separators.
- Capability inventory: The code performs LLM validation but does not exhibit dangerous capabilities such as file system writes, subprocess execution, or unauthorized network operations.
- Sanitization: Absent; the input text is not escaped or validated before being sent to the LLM.
Audit Metadata