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:
userQueryinRAGChat.chatanduserMessageinchatWithToolsinreferences/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