scenario-playable-ads
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the installation of widely used Python packages from PyPI (Pillow, NumPy, SciPy, FontTools, Brotli, and Playwright) and the download of the Chromium browser. These are standard dependencies for high-quality media processing and automated testing.
- [COMMAND_EXECUTION]: Python scripts within the skill use the subprocess module to invoke trusted system tools including ffmpeg and ffprobe. These calls are used appropriately for media analysis, video frame extraction, and asset compression within the project workflow.
- [DYNAMIC_EXECUTION]: The skill generates self-contained JavaScript and HTML deliverables for various ad networks (e.g., Meta, Google Ads). It uses Playwright to execute these generated bundles in a headless browser to perform automated security and functionality validation, such as ensuring no external network requests are made during ad runtime.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user content, including video recordings and JSON metadata. The risk of malicious instructions within this data is mitigated by the skill's architecture, which uses the agent to perform deterministic analysis against pre-defined archetype guides rather than executing commands found within the input media.
Audit Metadata