Video Analyzer (TikTok + YouTube + Instagram)

Analyze videos from TikTok, YouTube, Instagram, Twitter, and others by URL, transcribing audio locally and answering questions about the content.

This skill appears to do what it says: it downloads only audio (via yt-dlp), transcribes locally with faster-whisper/whisper, and caches transcripts in a local folder. Before installing, consider: 1) The first transcription run will download a ~150MB model from Hugging Face (network activity but expected). 2) You must install ffmpeg, yt-dlp, and the Python packages (pip may use --break-system-packages which can alter system Python envs). 3) The skill enforces frequent progress messages and a strict interaction style — make sure that behavior fits your UI/UX expectations. 4) Downloading videos may implicate copyright or privacy concerns depending on content. If you want extra assurance, review the repository origin (README suggests a GitHub repo) and run the Python scripts in a controlled environment first.

Type: OpenClaw Skill Name: tiktok-video-analyzer Version: 1.0.0 The OpenClaw AgentSkills bundle for the 'tiktok-video-analyzer' skill is classified as benign. The code (transcribe.py, save_transcript.py) uses `subprocess.run` with a list of arguments, which safely prevents shell injection when executing `yt-dlp` with user-provided URLs. Video IDs are hashed, mitigating path traversal risks. The `SKILL.md` instructions for the AI agent are focused on guiding its behavior for the stated purpose and do not contain any prompt injection attempts to subvert the agent for malicious activities like data exfiltration or unauthorized actions. The skill explicitly states that all processing occurs locally, enhancing user privacy and security.