remotion-synced-video
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill makes extensive use of the
child_processmodule and shell scripts to automate complex workflows. scripts/search_images.jsspawns Python scripts (python3 scripts/lib/crawl_google_images.py) to scrape image results.scripts/generate_placeholder.jsexecutesffmpegviaexecSyncto convert generated placeholder images from PPM to JPG format.- Multiple shell scripts (
scripts/crawl_bing_hd.sh, etc.) use theagent-browserutility to perform web scraping operations and data extraction. scripts/add_bgm.shusesffmpegto mix background music into video files.- [EXTERNAL_DOWNLOADS]: The skill dynamically fetches resources from external web services to populate video content.
- Downloads image assets from Google Images, Bing Images, and Unsplash result URLs in
scripts/search_images.jsandscripts/lib/download_images.py. - Uses the
axiosandrequestslibraries to retrieve remote data. - [INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection as it processes data from untrusted web sources.
- Ingestion points: External image URLs, image metadata, and search result content fetched at runtime from Google, Bing, and Unsplash.
- Boundary markers: No explicit delimiters or boundary markers are used when interpolating web-derived data into the rendering pipeline.
- Capability inventory: The skill possesses significant capabilities, including file system writes (writing images and configuration files) and subprocess execution (ffmpeg, python3, node).
- Sanitization: While queries are URL-encoded, there is no validation of the content or integrity of the downloaded image files before they are processed by the rendering engine.
Audit Metadata