minoan-frontend-design
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The evaluation framework (eval/run_eval.py and eval/run_eval_f2.py) downloads necessary Python dependencies and installs Playwright browser binaries to automate the rendering and screenshotting of designs.
- [COMMAND_EXECUTION]: The provided evaluation scripts execute shell commands to manage headless browser instances and process file operations as part of the design benchmarking pipeline.
- [DATA_EXPOSURE]: The evaluation tools access API keys for Anthropic and OpenRouter from the user's environment or a configuration file (~/.config/env/global.env). This is a legitimate requirement for the script to perform automated design assessments using LLMs.
- [DYNAMIC_EXECUTION]: eval/run_eval_f2.py utilizes the claude_agent_sdk to programmatically trigger agent sessions for the purpose of testing how different design instructions affect the output quality of various models.
Audit Metadata