io.github.CSOAI-ORG/video-editing-ai-mcp

Analyze video metadata and produce scene split points with timestamps and frame numbers. Useful for automatic chapter markers or editing cuts. Args: duration_seconds: Total video duration in seconds scene_type: Content type (dialogue, action, establishing, montage, transition) fps: Frames per second of the source video (default: 24) sensitivity: Detection sensitivity 0.1-1.0 (higher = more cuts)

Generate timed subtitles (SRT format) from a transcript. Automatically splits text into readable chunks with proper timing. Args: transcript: Full text transcript of the video duration_seconds: Total video duration in seconds style: Subtitle style (standard, bold, minimal, karaoke) max_chars_per_line: Maximum characters per subtitle line (default: 42)

Generate thumbnail selection recommendations including best timestamps, composition tips, and text overlay configuration. Args: title: Video title (used for text overlay recommendations) duration_seconds: Total video duration in seconds scene_count: Number of distinct scenes in the video style: Video style (gaming, tutorial, vlog, review, news)

Recommend color grading settings based on genre, mood, and lighting conditions. Returns contrast, saturation, temperature, LUT suggestions, and color wheel settings. Args: genre: Video genre (cinematic, vintage, documentary, horror, comedy, music_video, corporate, noir) mood: Emotional tone (happy, sad, tense, romantic, mysterious, energetic) lighting: Lighting condition (natural, studio, low_light, harsh, golden_hour, mixed) custom_adjustments: Optional overrides as {setting: value}

Plan an aspect ratio conversion with detailed crop/pad calculations, content preservation percentage, and FFmpeg command hints. Args: source_ratio: Source aspect ratio label (e.g. "16:9") target_ratio: Target aspect ratio (16:9, 9:16, 1:1, 4:3, 21:9, 4:5, 2.39:1) source_width: Source video width in pixels source_height: Source video height in pixels strategy: Conversion strategy (crop, letterbox, smart_crop, stretch, fill_blur)