kami-conflict-detection

• Analyzed: May 13, 2026, 6:56 AM
• Findings: 1
• Version: 1.0.1
• Hash: 0a152e7a95ef9c0d23aedbf9e3e8845a45ca98751c170a852df01acc549bae28
• Source: palm
• External report: View on VirusTotal

Findings (1)

The skill provides physical conflict detection by processing video streams with YOLO and a remote LLM API. It is classified as suspicious due to several high-risk capabilities: the 'conflict_detector_last.py' script automatically downloads model weights from GitHub and uses 'subprocess' to install Python dependencies (ultralytics, onnx) at runtime. Additionally, it transmits video frames to an external API (kamiclaw-skill-api.kamihome.com) and supports external notifications via Feishu webhooks. While these actions are consistent with the stated purpose of a smart-home security tool, the combination of runtime environment modification and remote data transmission meets the threshold for risky behavior without clear evidence of malice.