skills/pavelsimo/skills/ytd/Gen Agent Trust Hub

ytd

Pass

Audited by Gen Agent Trust Hub on Aug 3, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill executes a local Python script (ytd.py) through the uv tool. It uses PEP 723 inline metadata to declare its dependency on the yt-dlp library, ensuring a self-contained and verifiable execution environment.
  • [EXTERNAL_DOWNLOADS]: The script connects to YouTube domains to fetch video metadata, media streams, and subtitles. These network operations are strictly limited to the skill's documented purpose of media retrieval from a well-known service.
  • [COMMAND_EXECUTION]: The agent is instructed to invoke the ytd.py script via a shell command. The logic includes parsing user-supplied flags and URLs, following standard command-line interface patterns.
  • [PROMPT_INJECTION]: The skill processes transcripts retrieved from external YouTube URLs. Although this represents an indirect prompt injection surface, the script implements sanitization by removing HTML tags and formatting metadata, and the risk is considered low and inherent to the media processing use case.
  • [DATA_EXFILTRATION]: No evidence of unauthorized data exfiltration was found. The skill only accesses and saves the specific media content or transcripts requested by the user to the local filesystem.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 3, 2026, 04:14 PM
Security Audit — agent-trust-hub — ytd