copy

内容来源:README.md(说明文档) · 原始地址 · 查看安装指南

原始内容

HIGH ERA | Automated Creative Infrastructure

Established 2026.

High Era is the operating system for modern marketing agencies. It replaces the chaos of scattered briefs, lost context, and manual asset generation with a unified, intelligent "Agency Memory."

High Era Terminal

Two Ways to Run High Era

We believe in the power of open source and the convenience of managed services. Choose the path that fits your ambition.

1. The Managed Studio (Recommended)

For Marketing Directors, Agencies, and Teams.

Skip the infrastructure setup. Get instant access to the full High Era suite, including our premium Twin-Engine video generation, priority processing, and white-glove support.

  • Instant Access: Start briefing in 30 seconds.
  • Managed Infrastructure: We handle the GPUs, queues, and uptime.
  • Enterprise Security: SOC2 compliant (roadmap), encrypted Dossiers.
  • Zero Maintenance: Continuous updates and new skills automatically added.

→ Start Your Studio ($297/mo) | View Pricing

2. The Open Source Core

For Engineers, Hackers, and Builders.

Run the core engine on your own infrastructure. You have full control over the code, the models, and the data. You are the architect.

  • 🛠 Full Control: Modify the skills, tweak the prompts, own the stack.
  • 🛠 Self-Hosted: Run on your own GCP project or local machine.
  • 🛠 Community Driven: Contribute to the skill library.

→ View Deployment Guide


Key Capabilities

📂 The Dossier (Contextual Memory)

Stop re-explaining your brand. High Era remembers your products, audiences, and value propositions.

  • Persists Context: Briefs are saved and indexed.
  • Auto-Priming: The AI knows your voice before you write a word.

⚡️ Twin-Engine Workflow (Async Generation)

Don't wait for the progress bar.

  • Engine 1 (Synchronous): Instant strategy, copy, and ideas.
  • Engine 2 (Asynchronous): Background workers handle heavy lifting (Video rendering, Deep Research).
  • Notification System: Queue 10 campaigns, get notified when they're done.

🧠 The Skill Library (Madison Avenue Logic)

We don't just "prompt." We execute proven frameworks.

  • 25+ Specialized Skills: From "Hemingway-style Copy" to "Kling Video Scripts".
  • Rigorous Checklists: Every output is self-evaluated before you see it.

Local Development (Self-Hosted)

Prerequisites

  • Python 3.10+
  • Node.js 18+
  • Google Cloud Project (Firestore, Pub/Sub, Storage)
  • Anthropic API Key
  • FAL.ai Key (for image/video/audio generation with turbo models)

Quick Start

  1. Clone the Repo

    git clone https://github.com/high-era/core.git
    cd high-era
    
  2. Configure Environment

    cp .env.example .env
    # Add your API keys to .env:
    # - ANTHROPIC_API_KEY
    # - FAL_KEY (for asset generation)
    
  3. Start Backend (FastAPI)

    pip install -r requirements.txt
    python3 -m uvicorn service.main:app --reload --port 8080
    
  4. Start Frontend (SvelteKit)

    cd frontend
    npm install
    npm run dev
    
  5. Open the Terminal Visit http://localhost:3000 to access your local High Era instance.

  6. Generate Campaign Assets (Optional)

    # Generate hero images and campaign assets using FAL turbo models
    python scripts/generate_campaign_assets.py --all
    
    # Or use the Asset Assessment Lab at http://localhost:3000/assess
    

Asset Generation with Turbo Models

High Era includes integrated asset generation using FAL.ai's turbo models for ultra-fast, high-quality images, videos, and audio.

Available Turbo Models

Image Generation:

  • FLUX Schnell - Ultra-fast (2-4s), high-quality photorealism
  • SDXL Lightning - Ultra-fast (1-2s) for rapid iteration
  • Qwen Image 2512 - Fast with exceptional text rendering
  • FLUX Pro 1.1 - Premium photorealism for hero images
  • Recraft V3 - Design and vector-focused assets

Video Generation:

  • LTX Video - Fast video generation
  • Kling V2.5 Turbo Pro - High-quality turbo video

Audio Generation:

  • Stable Audio - Music and soundscapes

Generate Assets

Via UI:

  1. Visit /assess (Asset Assessment Lab)
  2. Select asset type (image/video/audio)
  3. Choose a turbo model
  4. Enter your creative prompt
  5. Generate and review

Via Script:

# Generate all assets for all audiences
python scripts/generate_campaign_assets.py --all

# Generate specific audience assets
python scripts/generate_campaign_assets.py --audience founders --type og

# Generate ad creatives
python scripts/generate_campaign_assets.py --audience freelancers --type ads

Via API:

curl -X POST http://localhost:8000/generate-asset \
  -H "Content-Type: application/json" \
  -d '{"type":"image","prompt":"Cinematic hero image","model":"fal-ai/flux/schnell"}'

See Asset Generation Guide for detailed documentation.

Deployment

The system is designed for Google Cloud Run.

# Deploy Backend
gcloud run deploy high-era-api --source .

# Deploy Frontend
cd frontend
gcloud run deploy high-era-ui --source .

See CONTRIBUTING.md for detailed architecture documentation.


License

High Era is open-source software licensed under the MIT license. You are free to use, modify, and distribute this software for private or commercial use.