Opus Orchestrator AI

A comprehensive AI-powered book generation system with LangGraph, CrewAI, and AutoGen.

🎯 What is Opus?

Opus is an AI-powered book generation system that creates professional manuscripts. It supports fiction and nonfiction with intelligent purpose classification, multiple export formats, and professional publishing workflows.

✨ Features

Core Capabilities

• Multi-Framework Orchestration: LangGraph, CrewAI, and AutoGen
• Intelligent Purpose Classification: Automatically determines reader purpose
• 100+ Content Frameworks: From textbooks to RPG modules
• Checkpoint/Resume: Long generations can resume from failure

Input Sources

• GitHub Repository: Ingest from any public/private repo
• S3/Backblaze: Cloud storage ingestion
• Local Files: Direct file input

Output Formats

• Scrivener Export: Chapter-by-chapter with binder.json
• LaTeX: 31 professional templates
• HTML: Styled web output
• PDF: Via LaTeX compilation

Publishing

• KDP Templates: 5 trim sizes (5x8, 5.5x8.5, 6x9, 8x8, 8.5x11)
• 31 LaTeX Templates: Novel, memoir, academic, RPG, cookbook, etc.
• ISBN/Metadata: Full publishing metadata support

Deployment

• Docker Compose: Full local stack
• k3s/Helm: Production Kubernetes
• REST API: Programmatic access

🚀 Quick Start

# Install
pip install opus-orchestrator

# Generate a book (fiction)
opus generate --concept "A robot who dreams of being human" --genre sci-fi

# Generate a book (nonfiction)
opus generate --book-type nonfiction --purpose learn "How to build an AI app"

# Serve API
opus serve --port 8000

📚 Framework Library

Opus supports 100+ frameworks organized by content type:

Nonfiction

Category	Frameworks	Purpose
Tutorial/How-To	Tutorial, Howto, Minimalist How-To	Learn to do
Explanation	Concept Explainer, Socratic Method	Understand
Transformation	Transformation Journey, Atomic Habits	Personal change
Decision	Big Idea, Case Study	Make decisions
Reference	Technical Manual, Quick Reference	Look up info

Fiction

Category	Frameworks
Snowflake	One-page to novel
Three-Act	Classic structure
Hero's Journey	Mythic structure
Save the Cat	Screenwriting
Story Circle	Dan Harmon's 8-part

RPG/Game Books

Category	Frameworks
Rulebook	Core rules, system
Adventure	Dungeon module, campaign
CYOA	Choose Your Own Adventure

📄 Output Formats

Scrivener Export

from opus_orchestrator import export_to_scrivener

result = export_to_scrivener(
    manuscript,
    "My Book",
    split_chapters=True,
    branch="draft/chapter-1",
    push_to_remote=True,
)

Output: Individual .md files + binder.json

LaTeX Templates (31)

from opus_orchestrator import export_to_latex

export_to_latex(manuscript, "My Book", "out.tex", 
    template="kdp-trade")

Templates:

• KDP: pocket, trade, 6x9, square, large
• Genre: novel, memoir, romance, thriller, sci-fi
• Academic: textbook, academic, cleanthesis, classicthesis
• Specialty: poetry, cookbook, screenplay, RPG

HTML Export

from opus_orchestrator import export_to_html

html = export_to_html(manuscript, "My Book", 
    template="memoir")

🏭 Deployment

Docker Compose

# Quick start
cp .env.example .env
# Add your MINIMAX_API_KEY and GITHUB_TOKEN

docker-compose -f deployments/docker-compose.yml up -d

k3s/Helm

# Install Opus API
helm install opus deployments/k3s/opus-orchestrator/

# Install TeX Live API (for PDF compilation)
helm install texlive deployments/k3s/texlive-api/

🔌 REST API

# Start server
opus serve

# Generate (blocking)
curl -X POST http://localhost:8000/generate \
  -H "Content-Type: application/json" \
  -d '{"concept": "Your book idea", "book_type": "fiction"}'

# Generate (streaming)
curl -X POST http://localhost:8000/generate/stream \
  -H "Content-Type: application/json" \
  -d '{"concept": "Your book idea"}'

⚙️ Configuration

Environment Variables

export MINIMAX_API_KEY="your-key"  # Primary
export GITHUB_TOKEN="your-token"

Config File

agent:
  provider: minimax
  model: MiniMax/MiniMax-M2.5
  temperature: 0.7

output:
  format: markdown
  save_to_file: true
  split_chapters: true

🧠 Architecture

User Input → Intent Classification → Framework Selection
    ↓
Purpose Detection (learn/understand/transform/decide)
    ↓
Framework匹配 (100+ frameworks)
    ↓
Agent Selection (purpose-specific)
    ↓
Generation (LangGraph/CrewAI/AutoGen)
    ↓
Output (Scrivener/LaTeX/HTML/PDF)

📁 Project Structure

opus_orchestrator/
├── orchestrator.py          # Main orchestration
├── server.py               # REST API
├── cli.py                  # CLI
├── langgraph_workflow.py   # LangGraph pipeline
├── nonfiction/             # Nonfiction system
│   ├── classifier.py       # Purpose classifier
│   └── frameworks.py       # 35+ frameworks
├── frameworks.py           # Fiction frameworks
├── scrivener_export.py    # Scrivener output
├── latex_compile.py        # LaTeX export
├── html_export.py          # HTML output
├── texlive_client.py       # TeX Live API
├── templates/
│   └── latex/             # 31 templates
└── deployments/
    ├── docker-compose.yml
    └── k3s/

📦 Test Suite

# Run tests
pytest tests/ -v

# Test categories
tests/
├── test_github_ingest.py   # GitHub ingestion
├── test_s3_ingest.py      # S3/Backblaze
├── test_generation.py     # Document generation
├── test_output_push.py    # Output push
└── test_e2e.py           # End-to-end

🔗 Links

• GitHub
• Issues
• BECOMINGONE

📜 License

MIT

👤 Author

Mark Havens

---

Built with ♥ using MiniMax M2.5