llm-app-patterns

Pass

Audited by Gen Agent Trust Hub on Apr 8, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [SAFE]: All 19 files are informational markdown documents. The skill does not include any executable scripts, binary files, or commands that interact with the host system.
  • [INDIRECT_PROMPT_INJECTION]: The RAG and agent patterns (e.g., in sub-skills/14-generation-with-context.md) demonstrate a design that interpolates external context into prompts. Ingestion points: Context documents and user-supplied variables. Boundary markers: Simple text labels like 'Context:' and 'Question:'. Capability inventory: Intended for LLM response generation. Sanitization: Not included in these structural code examples.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 8, 2026, 09:38 PM