youtube
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources, which could be used to influence the agent's behavior through malicious content embedded in transcripts or comments. -- Ingestion points: Untrusted data enters the context through execution/fetch_transcript.py (video subtitles) and sub-skills/competitor.md (audience comments mined via the DataForSEO tool). -- Boundary markers: Absent. The skill instructions do not require the use of delimiters or explicit instructions to ignore embedded commands when processing this external content. -- Capability inventory: The skill has access to powerful tools including Bash, Write, and Agent. It also includes custom Python scripts capable of making network requests and modifying local files. -- Sanitization: Absent. External content is processed as raw text without filtering or escaping.
- [COMMAND_EXECUTION]: The skill relies on local command execution to perform certain tasks. -- Evidence: The script execution/fetch_transcript.py executes the yt-dlp binary using subprocess.run to extract subtitles. Although it uses a list-based argument structure to prevent shell injection, it represents a capability to run external processes on the underlying system.
Audit Metadata