yt-viral
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The Python script
swipe.pyaccepts a file path as a command-line argument without validation. This allows the AI agent to potentially read any JSON-formatted file on the local system if the file path is manipulated. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted YouTube video titles and instructs the AI to summarize their shared structural elements. This creates an attack surface where a title containing malicious instructions could influence the AI's analysis results.
- Ingestion points: Video titles are read from the
collected.jsonfile inswipe.py. - Boundary markers: No delimiters or safety instructions are used to isolate the data from the agent's logic during the summary phase.
- Capability inventory: The script performs local file reads; the agent then processes the script's output to generate a response.
- Sanitization: The skill does not sanitize or escape the content of video titles before processing them.
Audit Metadata