Creative Engine
Creative Engine is a unified Claude Code skill pack for AI video and content production. It orchestrates 14 sub-skills across five categories — generation, API providers, post-processing, distribution, and discovery — through dual-mode routing (auto-detection of user intent, or direct commands), chain templates for end-to-end workflows, and session context for cross-skill artifact passing.
Version: 1.1.1 | License: MIT | Author: raphael
Skill Catalog
Generation
| Skill | Description |
|---|---|
| Cinematography | Design shots, scenes, aesthetics, transitions, and video prompts using cinematic language. 10-dimension interactive shot design protocol. |
| Storyboard | Convert stories and scripts into structured shot lists with character identity keywords for multi-shot consistency. |
| Character Bank | Manage a character asset library. Search, create, repair, and expand character JSON definitions. Generate multi-view grids and expression sheets. |
| Content Ops | Build repeatable AI content production systems: briefs, quality checklists, model selection, UGC scripts, and content strategy. |
| Production Line | Pipeline protocol library with 5 end-to-end production workflows (text-to-video, short-video, video-replication, batch-production, commercial). |
| Video Intel | Analyze, reverse-engineer, and download videos. Viral factor analysis, shot-by-shot breakdowns, and multi-platform video fetching. |
API Providers
| Provider | Description |
|---|---|
| Seedance (Jimeng) | Image + video generation via Seedance 2.0. Text-to-image, image-to-image, text-to-video, image-to-video, and transition video. |
| Kling | Video generation with multi-shot narrative, sound generation, and omni-video editing capabilities. |
| Sora | OpenAI Sora 2 text-to-video and image-to-video with auto-generated synced audio. Two models: sora-2 (fast) and sora-2-pro (quality). |
| Gemini | Google Gemini via flow4api. Video generation (Veo 3) and image generation (Nano Banana). Synchronous responses. |
Post-Process
| Skill | Description |
|---|---|
| Video Overlay | Overlay images and text on videos: logos, watermarks, subtitles, captions, stickers, brand marks. Auto-calculates position and size. |
| Video Assembler | Batch-assemble short videos from raw footage. Fixed opening/closing clips with permuted middle segments. |
| Audio Toolkit | Extract, trim, concat, mix, speed-change, and replace audio tracks. Includes TTS voice generation with emotion control. |
Distribution
| Skill | Description |
|---|---|
| IG Poster | Publish Instagram Reels with caption, thumbnail, location, and mentions. Supports scheduled posts, multi-account rotation, and performance tracking. |
Discovery
| Skill | Description |
|---|---|
| IG Hunter | Discover viral Instagram Reels from small accounts. Explore feed scanning, batch URL checking, and creator profile lookup. |
Installation
creative-engine installSee Creative Engine CLI for full setup instructions.
Requirements
- Claude Code or similar AI assistant with file reading capability
- uv and Python 3.10+ (for scripts)
- ffmpeg and ffprobe in PATH (for post-process skills)
- Dependencies are declared inline via PEP 723 and auto-installed by uv
How It Works
Creative Engine uses dual-mode routing:
- Auto mode — Analyzes user intent from natural language and recommends the appropriate sub-skill or chain template.
- Direct mode — Users explicitly activate sub-skills with trigger phrases (e.g., “use cinematography”, “use video-intel”).
Chain templates define end-to-end workflows that sequence multiple sub-skills. For example, the text-to-video chain runs storyboard, then cinematography, then the API provider. Session context tracks artifacts across skills so that character keywords, shot lists, and generated assets flow automatically between steps.