qinghu-viral-video-toolkit
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches Node.js binaries from the official nodejs.org distribution site and the @iqinghu/qhkit utility from the npm registry (or the Aliyun mirror registry.npmmirror.com) during its environment setup phase.
- [COMMAND_EXECUTION]: The instructions involve executing system commands including curl, tar, and ffmpeg, as well as the vendor-specific qhkit CLI tool to process video and audio data. An automated scanner flagged a checksum verification step (sha256sum -c) as a remote execution pipe; however, manual inspection confirms this is a security best practice for verifying the integrity of the downloaded Node.js binary, not a malicious code execution pattern.
- [PERSISTENCE]: The skill provides instructions to modify the user's ~/.bashrc file to ensure the locally installed Node.js binary path remains in the shell's PATH across sessions.
- [INDIRECT_PROMPT_INJECTION]: The skill processes content from external video links (e.g., Douyin or TikTok), creating an ingestion surface for potentially untrusted data.
- Ingestion points: The resourceUrl parameter in the qhkit video-inspire command, which fetches metadata from external URLs.
- Boundary markers: None identified; extracted scripts are used to generate new content directly.
- Capability inventory: Binary execution of qhkit and ffmpeg.
- Sanitization: No explicit sanitization of the remote video metadata or script content is detailed in the skill instructions.
Audit Metadata