Video Content Analyzer

下载视频并用AI分析内容 - 支持B站/抖音/YouTube等平台，提取语音内容并分析视频结构

This skill appears to do what it says (download videos, extract audio, transcribe with Whisper, analyze frames), but the manifest omitted that it needs an OpenAI API key and may read/write ~/.openclaw/openclaw.json. Before installing or running: 1) confirm you are comfortable granting Exec permission to run yt-dlp/ffmpeg and downloading arbitrary URLs (copyright risk); 2) do not reuse a high‑privilege OpenAI key — create a limited/test key or set usage limits; 3) avoid running pip with --break-system-packages on critical systems; 4) audit any secret stored in ~/.openclaw/openclaw.json and consider keeping it in an isolated environment; 5) ask the publisher to update metadata to declare OPENAI_API_KEY and the config path so the required permissions are explicit. If you need higher assurance, run the workflow manually in a sandbox first.

Type: OpenClaw Skill Name: video-content-analyzer Version: 0.1.0 The skill utilizes the 'Exec' tool to perform video downloading and processing via yt-dlp and ffmpeg, which are high-risk capabilities involving shell execution and network access. While these actions are aligned with the stated purpose of video content analysis and no malicious intent or data exfiltration was found, the broad permissions and execution of external binaries (SKILL.md) warrant a suspicious classification per the defined criteria.