Youtube Clipper Ko

한국어 YouTube 영상을 바이럴 클립으로 자동 분할하는 스킬. Whisper로 음성 분석 → Claude로 바이럴 구간 선정 → ffmpeg으로 클립 추출. 무음/숨소리 구간 제거, 3색 한국어 자막(SUIT 폰트), 바이럴 점수 + 리즌 텍스트 포함. YouTube URL 또...

Before installing, be aware of these points: - Privacy/hosting: Despite README claims of 'local-only', the script uploads audio to OpenAI Whisper and sends transcripts/prompts to Anthropic for analysis. If you must keep audio/video entirely offline, do not use this skill as-is (consider replacing Whisper/Anthropic calls with a local ASR/LLM). - Credential visibility: The code requires OPENAI_API_KEY and an Anthropic API key (used at runtime). The registry metadata did not declare these env vars — verify and set keys only after you accept the external API usage. - Data sensitivity & copyright: Using yt-dlp downloads remote videos; uploading audio to external APIs may violate privacy or copyright expectations. Only process content you are allowed to upload to those services. - Confirm author/source: The README points to a GitHub repo and an author handle; if you care about provenance, inspect the repo and commits. Prefer installing from a known repository rather than an anonymous bundle. - Running safely: Run with the --dry option first to inspect transcripts and outputs. Consider running in an isolated environment (container, VM) and limit network egress if you want to inspect behavior before allowing API calls. Rotate keys after testing and scope them appropriately. - Ask for fixes: Request the publisher update the registry metadata to declare required env vars (OPENAI_API_KEY, ANTHROPIC_API_KEY) and correct the 'local-only' wording in README/SKILL.md so it accurately reflects external API usage. If you want a fully offline alternative, ask the author or maintainers whether the tool supports a local-only ASR (e.g., whisper.cpp) and local LLM processing, or consider forking the project and replacing network calls.

Type: OpenClaw Skill Name: youtube-clipper-ko Version: 0.1.0 The youtube-clipper-ko skill is a legitimate video processing tool designed to automate the creation of short-form clips from YouTube videos. It utilizes standard libraries and tools such as yt-dlp for downloading, OpenAI Whisper for transcription, and Anthropic Claude for content analysis. The Python script (scripts/clip.py) safely handles external process execution using subprocess lists to prevent shell injection, and the instructions in SKILL.md are strictly aligned with the tool's stated purpose without any evidence of malicious prompt injection or data exfiltration.