youtube-dl
Pass
Audited by Gen Agent Trust Hub on Mar 1, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The script
scripts/download.pyautomatically installs theyt-dlppackage from PyPI usingpip installif it is not already present on the system. Whileyt-dlpis a well-known tool, automatic package installation is a noteworthy behavior. - [COMMAND_EXECUTION]: The skill executes
python3to run its bundleddownload.pyscript. The script itself usessubprocess.runto execute thepipcommand for dependency management. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8). It fetches and displays untrusted metadata (titles, descriptions, uploader names) from external video platforms.
- Ingestion points: Metadata is ingested via
ydl.extract_info(url)inscripts/download.pyand printed to the console. - Boundary markers: No boundary markers or 'ignore' instructions are used when printing the external metadata to the agent's context.
- Capability inventory: The skill has the capability to write files to the local filesystem (video downloads) and execute shell commands (via
subprocess.runduring setup). - Sanitization: The script truncates descriptions to 200 characters but does not perform sanitization or escaping of the content to prevent instructions embedded in video titles or descriptions from influencing the agent's next steps.
Audit Metadata