mistral-core-workflow-b

Pass

Audited by Gen Agent Trust Hub on Mar 24, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: No security issues detected. The skill implements standard Mistral AI SDK patterns for text embeddings and function calling.
  • [PROMPT_INJECTION]: The skill processes user-supplied text for search and RAG workflows in references/implementation.md. This represents an indirect prompt injection surface common to retrieval-augmented generation systems.
  • Ingestion points: userQuery in RAGChat.chat and userMessage in chatWithTools in references/implementation.md.
  • Boundary markers: The RAG implementation uses system prompt instructions ("Answer based on context. If not in context, say so.") and clear context headers to delineate untrusted data.
  • Capability inventory: No dangerous capabilities such as file writes or shell command execution are utilized within the provided implementation scripts.
  • Sanitization: External input is interpolated directly into prompts without explicit sanitization or escaping, which is standard for LLM integration logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 24, 2026, 05:03 PM