axiom-foundation-models
Pass
Audited by Gen Agent Trust Hub on Mar 5, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is composed of documentation and non-executable Swift code snippets. It does not perform any automated tasks, install dependencies, or execute system commands.
- [PROMPT_INJECTION]: The documentation provides explicit warnings against unsafe practices like interpolating user input into system instructions. It encourages the use of structured output (@Generable) and framework-level guardrails to ensure model safety.
- [DATA_EXFILTRATION]: The skill advocates for local, on-device processing of data to maintain user privacy and avoid the security risks associated with sending sensitive information to third-party cloud LLM providers.
Audit Metadata