qiaomu-wx-video

Warn

Audited by Gen Agent Trust Hub on Jul 23, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads an external binary from the 'ltaoo/wx_channels_download' GitHub repository during the installation process via 'scripts/install_backend.py'.
  • Evidence: The script 'scripts/install_backend.py' fetches zip/tar.gz assets from GitHub releases based on a pinned 'release-lock.json' file.
  • Mitigation: The skill performs SHA-256 checksum verification and requires explicit user consent ('--accept-upstream-license') before downloading.
  • [COMMAND_EXECUTION]: The skill executes the downloaded 'wx_video_download' binary to perform video capture and interacts with its API.
  • Evidence: Referenced in 'references/workflow.md' and the 'SKILL.md' workflow steps, including setting execution bits via 'chmod' in 'scripts/install_backend.py'.
  • [DATA_EXFILTRATION]: The skill can send user-provided WeChat Channels share links to a third-party online resolution service ('https://sph.litao.workers.dev/').
  • Evidence: Described in 'references/workflow.md' and 'SKILL.md' as the preferred 'no-install' path.
  • Mitigation: The skill requires separate user consent before sending the URL to the third-party service and explicitly forbids sending cookies or login states.
  • [COMMAND_EXECUTION]: The skill modifies system-wide network settings, including HTTP/HTTPS/SOCKS proxy configurations and root certificate installation.
  • Evidence: Described in 'references/security.md' and 'references/workflow.md' as necessary for capturing encrypted traffic.
  • Mitigation: The skill mandates individual consent for root certificate and proxy changes and includes a recovery/rollback phase to restore proxy settings from a snapshot.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 23, 2026, 02:02 AM