muapi-youtube-shorts

Pass

Audited by Gen Agent Trust Hub on Sep 30, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill relies on the execution of a local bash script (library/social/youtube-shorts/scripts/run-youtube-shorts.sh) and system utilities like jq and xargs to process video data.
  • [EXTERNAL_DOWNLOADS]: The skill interacts with the muapi.ai managed endpoint (/ai-clipping) for processing and references a GitHub repository (SamurAIGPT/AI-Youtube-Shorts-Generator) as a reference implementation. It also involves uploading local files to the muapi CDN.
  • [INDIRECT_PROMPT_INJECTION]: The skill presents a command injection surface by interpolating user-provided data (e.g., --source "<YOUTUBE_URL>") directly into shell commands.
  • Ingestion points: The --source input parameter in SKILL.md accepts arbitrary URLs or local file paths.
  • Boundary markers: None are defined to separate user input from the shell command structure.
  • Capability inventory: The skill uses bash to execute scripts and uses the muapi-cli tool to interact with remote services, including file uploads.
  • Sanitization: There are no instructions for the agent to sanitize or validate the format of the source URL or file path before execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 30, 2026, 07:17 AM