regex-vs-llm-structured-text
Pass
Audited by Gen Agent Trust Hub on Aug 5, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill implements a pipeline that processes untrusted text through an LLM, creating an indirect prompt injection surface.
- Ingestion points: The
process_documentandvalidate_with_llmfunctions inSKILL.mdaccept arbitrary string content from external documents. - Boundary markers: Absent; the prompt template in
validate_with_llminterpolatesoriginal_textdirectly into the user message without using XML tags, triple quotes, or other delimiters to isolate the data from the instructions. - Capability inventory: The skill is limited to returning structured data objects (
ParsedItem). It does not have access to the file system, network tools, or shell command execution. - Sanitization: No sanitization or safety filtering is applied to the input text before it is sent to the LLM validator.
- [SAFE]: No other security concerns were identified. The logic is deterministic, uses standard libraries, and contains no hidden or obfuscated code. The described approach follows legitimate architectural patterns for optimizing LLM costs.
Audit Metadata