data-engineering-ai-ml

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [Category 8: Indirect Prompt Injection] (SAFE): The skill implements Retrieval-Augmented Generation (RAG) patterns, which inherently involve processing external data into LLM prompts. This surface is managed through appropriate boundary markers.
  • Ingestion points: Documents are ingested from local Parquet files, LanceDB tables, and CSV files as seen in SKILL.md and rag-pipelines.md.
  • Boundary markers: Prompt templates in rag-pipelines.md use delimiters like '---' and structured tags such as '<|system|>' and 'Context:' to separate data from instructions.
  • Capability inventory: The skill performs local file/database operations and network requests to standard AI services (OpenAI). It does not contain subprocess execution or dynamic code evaluation (eval/exec).
  • Sanitization: Content is used as-is for semantic search and context assembly, which is standard for the described RAG use-case.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:31 PM