NYC
skills/astronomer/agents/airflow/Gen Agent Trust Hub

airflow

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
  • REMOTE_CODE_EXECUTION (HIGH): The hook script hooks/warm-uvx-cache.sh triggers a uvx command to download and execute astro-airflow-mcp@latest automatically upon SessionStart. This pattern runs remote code on the host machine without explicit user consent for each execution.
  • EXTERNAL_DOWNLOADS (HIGH): The skill relies on uvx to fetch the astro-airflow-mcp package from a non-trusted repository. As the package is not part of the defined [TRUST-SCOPE-RULE] organizations, its automatic execution is considered high risk.
  • COMMAND_EXECUTION (MEDIUM): The core functionality of the skill involves executing shell commands through uvx to manage Airflow workflows, which provides a significant capability surface on the local system.
  • CREDENTIALS_UNSAFE (LOW): The skill manages Airflow API tokens and passwords. While it supports environment variables, the documentation also promotes storing these secrets in a plaintext YAML configuration file at ~/.af/config.yaml.
  • DATA_EXPOSURE (LOW): The skill is designed to read and display sensitive Airflow data, including connections and variables, which may contain environment-specific secrets.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 17, 2026, 04:45 PM