Skills
skills/daviddwlee84/agent-skills/marimo-batch-mlflow/Gen Agent Trust Hub

marimo-batch-mlflow

Pass

Audited by Gen Agent Trust Hub on Apr 24, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references and uses the official huggingface_hub library to interact with Hugging Face, a well-known platform for machine learning model hosting and compute jobs.
  • [REMOTE_CODE_EXECUTION]: The grid search utility script (references/grid.py) uses huggingface_hub.run_uv_job to programmatically launch training tasks on Hugging Face. This is the primary intended functionality of the sweep launcher and utilizes established third-party integration patterns.
  • [COMMAND_EXECUTION]: The skill leverages Tyro for typed command-line argument parsing and employs shlex.quote in its grid launcher to safely handle shell arguments, adhering to security best practices for process execution.
  • [CREDENTIALS_UNSAFE]: Secret management is handled correctly; the skill instructs users to utilize .env files for sensitive tokens like HF_TOKEN and MLFLOW_TRACKING_URI rather than hardcoding credentials. It also uses the EnvConfig component for interactive secret entry in UI mode.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 24, 2026, 04:34 AM