| name | funclip-commander |
| displayName | Funclip Commander |
| description | Integrates OpenClaw with the FunClip AI video editor to enable AI-powered video recognition and clipping via command line. |
| version | 0.1.0 |
Funclip Commander
Description
Integrates OpenClaw with the FunClip AI video editor to enable AI-powered video recognition and clipping via command line.
Funclip Commander Skill
This skill allows OpenClaw agents to leverage the powerful features of the open-source FunClip AI video editor. By utilizing FunClip's command-line interface, agents can:
- Perform speech recognition on video files.
- Generate SRT subtitles.
- Clip specific segments of videos based on text prompts or speaker diarization.
Usage
This skill will expose functions to:
funclip.recognize_video(file_path_or_url: str, output_dir: str): Handles video download (if YouTube URL), extracts audio, and performs speech-to-text transcription. Depending on config.json settings, it uses Buzz (offline), OpenAI Whisper (cloud), or ClawClip's native ASR. Returns (video_filepath: str, audio_filepath: str, srt_filepath: str, srt_content: str).funclip.clip_video(file_path_or_url: str, output_dir: str, srt_file_path: str, dest_text: str = None, start_ost: int=0, end_ost: int=0, output_file: str = None): Clips video based on a provided SRT, recognized text, or timestamps using ClawClip's Stage 2.
Commands
This section would list the commands exposed by the OpenClaw skill to interact with FunClip.
Purpose
The Funclip Commander skill acts as a bridge between OpenClaw's agentic capabilities and FunClip's sophisticated video processing. It enables automated, AI-driven video content creation and analysis workflows directly from your OpenClaw environment.
Prerequisites
- FunClip Installed Locally: The FunClip repository (
https://github.com/modelscope/FunClip) must be cloned to your OpenClaw workspace within a directory named FunClip. All its Python dependencies must be installed in a virtual environment (FunClip/.venv).
ffmpeg and ffprobe: These must be available in the system's PATH.imagemagick: (Optional, for embedded subtitles) Must be installed and configured as per FunClip's README.md.yt-dlp: Required for downloading YouTube videos. Must be available in the system’s PATH.
Configuration (config.json)
- funclip_path: Path to the ClawClip repo (e.g.
workspace/ClawClip).
- venv_path: Path to ClawClip's virtual environment (e.g.
workspace/ClawClip/.venv).
- use_buzz_for_recognition:
true to use Buzz for ASR (default if key/path provided). Requires buzz_python.
- buzz_python: Full path to the Python executable within Buzz's virtual environment (e.g.,
/Users/ghost/.openclaw/workspace/ClawClip/buzz_venv/bin/python3). Buzz 1.4.x requires Python 3.12 or older.
- use_whisper_for_recognition:
true to use OpenAI Whisper API for ASR. Conflicts with use_buzz_for_recognition.
- whisper_model: Whisper model to use (default:
whisper-1). Requires OPENAI_API_KEY to be configured in Gateway.
Recognition Backends (selected via config, buzz has priority over whisper):
- Buzz (offline): Requires a Python 3.12 environment with Buzz installed. Ensures privacy and local processing. Configure
buzz_python and set use_buzz_for_recognition: true.
- OpenAI Whisper API (cloud): Requires an
OPENAI_API_KEY to be configured in the OpenClaw Gateway. Best for accuracy and convenience if API access is available. Set use_whisper_for_recognition: true.
- ClawClip Native (FunASR): Fallback to ClawClip's built-in ASR if neither Buzz nor Whisper are configured or available.
