pipeline-model-interpreter
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides instructions for interpreting pipeline models without requesting dangerous permissions, network access, or external code execution. All referenced files are local to the skill package.
- [DATA_EXPOSURE]: The skill identifies the potential for sensitive data within pipeline configurations and explicitly instructs the AI to redact or hide password fields and sensitive parameters, preventing accidental disclosure during interpretation.
- [INDIRECT_PROMPT_INJECTION]: As a data-processing skill, it handles external pipeline definitions which could potentially contain malicious prompts. The skill mitigates this risk by instructing the agent to remain conservative, avoid guessing the purpose of unknown plugin codes (atomCode), and follow a structured analysis playbook that focuses on structural metadata rather than executing content.
Audit Metadata