cap-llm-plugin

Pass

Audited by Gen Agent Trust Hub on Jul 29, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill recommends installing official SAP AI SDK packages (@sap-ai-sdk/langchain, @sap-ai-sdk/foundation-models) for Node.js in SKILL.md.
  • [PROMPT_INJECTION]: The code implementation in SKILL.md demonstrates the interpolation of user-provided content and retrieved document context directly into prompt templates.
  • Ingestion points: req.data in the onAskQuestion handler.
  • Boundary markers: Limited to simple text labels like Context: and Question: without escaping or explicit instructions to ignore embedded commands.
  • Capability inventory: The skill uses the SAP AI SDK to perform chat completions and vector database searches.
  • Sanitization: No sanitization or filtering logic is included in the examples, leaving the prompt vulnerable to indirect injection from processed data.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 29, 2026, 12:50 AM
Security Audit — agent-trust-hub — cap-llm-plugin