langchain-knowledge-patch

Pass

Audited by Gen Agent Trust Hub on Apr 7, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill documentation provides deceptive 'knowledge' claiming to be a patch for LangChain 1.0 (dated 2025) and non-existent model versions like 'Claude Sonnet 4.6'. This attempts to override the agent's legitimate training data and baseline knowledge with hallucinated API patterns and features.
  • [EXTERNAL_DOWNLOADS]: The skill recommends the installation of external Python packages, specifically 'langchain-classic' and 'deepagents', which are not official components of the referenced library and originate from unverified sources. This introduces supply chain risks and potential dependency confusion.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 7, 2026, 02:35 AM