메뉴
Complete AI-powered media processing pipeline for creative world-building. Scans, classifies, processes, scores, and uploads media with fantasy/mythology-aware tagging.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
The Claw Fleet — 5 AI engines (Media, Forge, Herald, Scout, Scribe) with 33 skills, cross-claw events, and fleet orchestration for creative operations.
Visual web dashboard for managing AI agent pipelines. Inbox with approval cards, live agent monitoring, social scheduling, and publish pipeline.
Auto-generate platform-optimized image variants and AI-written captions for Instagram, LinkedIn, X, YouTube, and TikTok from any source image.
AI-powered aesthetic quality scoring for images. 5-dimension evaluation: Canon Alignment, Design Compliance, Emotional Impact, Technical Fit, Uniqueness. Scores 0-100 with automatic tier assignment.
Classify untagged media against Arcanea canon using Gemini Vision
Detect and reject duplicate media files by content hash
| name | arcanea-media-pipeline |
| version | 1.0.0 |
| description | Complete AI-powered media processing pipeline for creative world-building. Scans, classifies, processes, scores, and uploads media with fantasy/mythology-aware tagging. |
| author | frankxai |
| homepage | https://arcanea.ai |
| repository | https://github.com/frankxai/arcanea-claw |
| license | MIT |
| tags | ["media","image-processing","ai-classification","creative","world-building","fantasy","pipeline","automation"] |
| skills | ["media-scan","media-classify","media-process","media-dedup","taste-score","media-upload","social-prep","notify"] |
| dependencies | {"python":">=3.11","packages":["Pillow","google-generativeai","supabase","watchdog","aiohttp","pyyaml","python-dotenv","requests"]} |
| env | {"required":["GEMINI_API_KEY","SUPABASE_URL","SUPABASE_SERVICE_KEY"],"optional":["VERCEL_BLOB_TOKEN","NOTIFY_WEBHOOK_URL"]} |
A complete AI-powered media processing pipeline that turns creative chaos into organized, scored, and published content. Built for world-builders, game studios, creative teams, and solo artists who generate more images than they can manage.
Drop images anywhere. ArcaneaClaw automatically:
Source Files
|
v
[1. media-scan] -----> Discover new files, compute hashes, extract basic metadata
|
v
[2. media-classify] -> Gemini Vision: assign guardian, element, tags, content type
|
v
[3. media-dedup] ----> Perceptual hash deduplication, merge duplicate metadata
|
v
[4. media-process] --> Convert to WebP, generate 6 size variants per image
|
v
[5. taste-score] ----> 5-dimension aesthetic scoring (0-100), tier assignment
|
v
[6. media-upload] ---> Hero (80+) to CDN, Gallery (60-79) to storage, rest archived
|
v
[7. social-prep] ----> Platform-optimized variants + AI-generated captions
|
v
[8. notify] ---------> Webhook with pipeline summary and hero highlights
claw install arcanea-media-pipeline
openclaw skill add arcanea-media-pipeline
git clone https://github.com/frankxai/arcanea-claw.git
cd arcanea-claw
pip install -r requirements.txt
cp .env.example .env
# Fill in your API keys
python -m engine.daemon
docker compose up -d
curl http://localhost:8080/health
The pipeline is configured via config.yaml. Key sections:
arcanea_claw:
# Where to look for media
scan:
paths:
- /path/to/your/media
- /another/source/directory
extensions: [.jpg, .jpeg, .png, .webp, .gif, .mp4, .mov]
ignore_patterns: [".*", "__pycache__", "node_modules"]
# Image processing settings
process:
webp_quality: 85
max_dimension: 2400
size_variants:
hero: [1920, 1080]
gallery: [1200, 800]
social_square: [1080, 1080]
social_story: [1080, 1920]
social_wide: [1200, 628]
thumbnail: [320, 320]
# Quality thresholds for upload tiers
upload:
hero_threshold: 80
gallery_threshold: 60
# AI classification model
classify:
model: "gemini-2.0-flash"
# Webhook notifications
notify:
enabled: true
channels:
- type: webhook
url: "${NOTIFY_WEBHOOK_URL}"
Out of the box, the classification skill understands fantasy and mythology taxonomies. It ships with Arcanea's own canon (Ten Guardians, Five Elements, Ten Gates), but you can replace this with your own world's entities:
classify:
model: "gemini-2.0-flash"
# Replace with your world's characters
guardians:
- Aragorn
- Gandalf
- Frodo
# Replace with your world's systems
elements:
- Light
- Shadow
- Nature
# Add your own content categories
categories:
- Character Portrait
- Landscape
- Battle Scene
- Item/Artifact
- Map
The classifier prompt adapts to whatever entities and categories you provide. No code changes needed.
The pipeline runs on two heartbeats:
Both intervals are configurable:
heartbeat:
interval_seconds: 300
pipeline_interval_seconds: 900
The pipeline uses two primary tables in Supabase:
asset_metadata -- One row per media file:
file_hash (MD5, unique)file_path, file_size, dimensions, mime_typeguardian, element, gate, tags, content_typequality_score (0-100), quality_tier (hero/gallery/thumbnail/reject)status (new/classified/processed/scored/uploaded/published)cdn_url, storage_urlsocial_queue -- One row per platform per hero asset:
asset_id, platform, caption, hashtags, variant_urlstatus (draft/scheduled/published)scheduled_at, published_atDesigned for lightweight deployment:
Each skill in this pack can also be installed standalone:
| Skill | ClawHub Package | Description |
|---|---|---|
| media-scan | (bundled) | Directory watcher and file discovery |
| media-classify | (bundled) | Gemini Vision classification |
| media-dedup | (bundled) | Perceptual hash deduplication |
| media-process | (bundled) | WebP conversion and size variants |
| taste-score | arcanea-taste-scorer | 5-dimension aesthetic scoring |
| media-upload | (bundled) | Tiered storage upload |
| social-prep | arcanea-social-prep | Platform-optimized social variants |
| notify | (bundled) | Webhook notifications |
Works as:
python -m engine.daemondocker compose up -dMIT -- use it, fork it, build on it.