video-audio-replace
ReviewAudited by ClawScan on May 1, 2026.
Overview
The artifacts consistently implement a user-run video dubbing tool, with expected but notable use of external TTS services, an optional ElevenLabs API key, local media processing tools, and unpinned setup dependencies.
This skill appears coherent for replacing video audio with TTS. Before installing or using it, make sure you are comfortable sending subtitle text to the selected TTS provider, use a dedicated ElevenLabs key if needed, install dependencies in a virtual environment rather than with sudo, and choose output filenames that will not overwrite important files.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The tool can overwrite the chosen output file and create temporary media files while processing videos.
The script invokes ffmpeg/ffprobe on local media paths and uses ffmpeg's overwrite flag. This is central to the stated video-audio replacement purpose, but it gives the skill local file-processing and output-file mutation capability.
subprocess.run(["ffmpeg", "-y", "-i", video_file, "-i", audio_file, ... output_file], capture_output=True)
Run it only on intended files, choose a fresh output filename, and keep backups of important media.
Using ElevenLabs can consume account quota or incur provider-side usage tied to the API key.
The ElevenLabs path reads an API key from the environment and sends it as the authentication header. This is expected for ElevenLabs TTS, but it grants the skill use of that TTS account.
api_key = os.environ.get("ELEVENLABS_API_KEY") ... "xi-api-key": api_keyUse a dedicated/revocable ElevenLabs key, monitor usage, and choose the Edge TTS mode if you do not want to provide an ElevenLabs key.
Private subtitle or transcript text may leave the local machine when using cloud/provider TTS.
Subtitle text is sent to an external TTS provider API for audio generation. This is purpose-aligned and disclosed by the skill's ElevenLabs/Edge TTS design, but it is still a data boundary users should notice.
data = {"text": text, "model_id": "eleven_multilingual_v2", ...}; response = requests.post(url, json=data, headers=headers)Avoid processing confidential transcripts unless the selected TTS provider's privacy and retention terms are acceptable.
Following that command could modify system Python packages and increase dependency supply-chain risk.
The setup guidance is not automatically executed, but it recommends a privileged, system-level pip install. The skill also relies on unpinned Python dependencies, so users should install cautiously.
print("Install with: sudo pip3 install faster-whisper --break-system-packages")Install dependencies in a virtual environment, avoid sudo/--break-system-packages when possible, and pin or verify package versions from trusted sources.