nlp-pipeline-builder

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • Indirect Prompt Injection (SAFE): The skill is designed to ingest and process external text data and CSV files, which is a common surface for indirect prompt injection. However, this is inherent to the NLP use case, and the skill provides tools for input sanitization.
  • Ingestion points: Processes user-provided text strings and datasets (train.csv, val.csv, test.csv) through the NLPPipeline class.
  • Boundary markers: No explicit delimiters are mentioned for the processed text, though the context is restricted to ML training and inference.
  • Capability inventory: Includes model training (fit), evaluation commands (/ml:nlp-evaluate), and deployment routines (/ml:nlp-deploy).
  • Sanitization: The TextPreprocessor class explicitly includes steps to remove HTML, URLs, and emails, which mitigates malicious content in data.
  • Unverifiable Dependencies (SAFE): The skill references the specweave Python library. While this is not a common public package, the skill only provides documentation on how to use it and does not contain commands to download or install code from untrusted remote sources.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:06 PM