natural-language-processing
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides standard, safe instructional content and code examples for NLP tasks using well-known and trusted libraries such as NLTK, scikit-learn, and Hugging Face Transformers.
- [INDIRECT_PROMPT_INJECTION]: The skill includes a data ingestion surface for indirect prompt injection via the processing of untrusted text data (e.g., in
preprocess_textin SKILL.md). However, the code includes basic sanitization (regex-based URL removal) and does not utilize any high-privilege tools, file system writes, or network operations, which limits the potential impact of adversarial content to a negligible level.
Audit Metadata