arize-prompt-optimization

Pass

Audited by Gen Agent Trust Hub on May 13, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides legitimate utility for prompt engineering and LLM observability.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: Includes explicit instructions (in SKILL.md and references/ax-profiles.md) to avoid reading sensitive local files like .env and to manage credentials securely via environment variables and official CLI profiles.
  • [EXTERNAL_DOWNLOADS]: References the installation of the 'arize-ax-cli' package from standard package registries (PyPI), which is the official tool for the service.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes production trace data which could contain malicious prompts. However, it uses clear boundary markers in its optimization meta-prompt and acts as a tool for a developer to analyze data rather than auto-executing untrusted content.
Audit Metadata
Risk Level
SAFE
Analyzed
May 13, 2026, 09:07 AM