ai-ml

Pass

Audited by Gen Agent Trust Hub on May 9, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill contains implementations for Retrieval-Augmented Generation (RAG) and text classification in SKILL.md that are vulnerable to indirect prompt injection.
  • Ingestion points: Untrusted data enters the agent context through the context, question, and userInput variables.
  • Boundary markers: The implementation uses basic string interpolation with newlines or quotes, which lacks the robust delimiters needed to prevent an attacker from escaping the data context to provide malicious instructions.
  • Capability inventory: The skill snippets demonstrate interactions with external LLM APIs (Anthropic, OpenAI) but do not execute shell commands or access sensitive local files.
  • Sanitization: No sanitization, escaping, or validation logic is applied to the untrusted inputs before they are interpolated into the prompt.
Audit Metadata
Risk Level
SAFE
Analyzed
May 9, 2026, 05:17 PM
Security Audit — agent-trust-hub — ai-ml