youtube-video-downloader
Pass
Audited by Gen Agent Trust Hub on Jul 24, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill sends network requests to
https://redfox.hk/story/api/parseWork/videoDownload/youtubeto parse video URLs. This domain is the official API endpoint for the vendor associated with the skill. - [COMMAND_EXECUTION]: The skill execution relies on a Python script (
scripts/downloader.py) that handles argument parsing, API interaction, and output formatting. - [PROMPT_INJECTION]: The skill presents a potential indirect prompt injection surface because it fetches and displays external video metadata (titles and descriptions) from the YouTube API response without using boundary markers or sanitization. While this is necessary for its primary purpose, malicious metadata could theoretically be crafted to influence agent behavior.
- Ingestion points: The
descfield in the API response is printed directly to stdout inscripts/downloader.py. - Boundary markers: Video descriptions are printed without delimiters or specific instructions to the agent to ignore embedded commands.
- Capability inventory: The skill environment allows for shell command execution and network access via Python scripts.
- Sanitization: There is no filtering or escaping of the retrieved description text before it is returned to the agent context.
Audit Metadata