mistral
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides implementation examples for Chat and Fill-in-the-Middle (FIM) completions. These patterns represent a vulnerability surface where instructions hidden in data could influence agent behavior.
- Ingestion points: Multiple code snippets in
SKILL.mddemonstrate ingesting strings intoclient.chat.completeandclient.fim.completemethods. - Boundary markers: The provided examples lack explicit boundary markers or delimiters to differentiate between system instructions and potentially untrusted external data (e.g., code prefixes or user messages).
- Capability inventory: The skill introduces tool-calling capabilities (e.g.,
lookup_dns_records), which allow the agent to perform actions based on model outputs derived from processed data. - Sanitization: There are no sanitization, validation, or escaping mechanisms shown for the data interpolated into model prompts.
Audit Metadata