wechat-channels-copy-extract
Pass
Audited by Gen Agent Trust Hub on Jul 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/extract_wechat_channels_copy.jsexecutes external commandsffmpeg,whisper, andnodeusingspawnSync. It passes arguments as an array, which mitigates shell injection risks. - [EXTERNAL_DOWNLOADS]: The skill depends on a sibling skill
wechat-channels-downloadto fetch video content from external WeChat servers via the TikHub API. - [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it transcribes audio from external videos without boundary markers. Malicious instructions spoken in a video could be included in the Markdown output, potentially influencing the behavior of an agent that processes the resulting file.
- Ingestion points: Video audio content (via Whisper), video metadata (via TikHub API), and user-supplied titles in
scripts/extract_wechat_channels_copy.js. - Boundary markers: None present in the prompt templates or output generation.
- Capability inventory: Subprocess execution of system tools (
ffmpeg,whisper) and other Node.js scripts. - Sanitization: Includes basic filename sanitization (
safeFileName) and HTML tag removal from metadata, but does not sanitize the transcript text for potential instructions.
Audit Metadata