Contract Playbook Enforcement Engine

Review incoming contracts against your negotiation playbook. Upload a contract, select a playbook, and get a deviation report with traffic-light risk ratings, fallback suggestions, and escalation flags.

---

Getting Started

Option 1: Run Locally (Recommended)

git clone https://github.com/your-org/contract-playbook-engine.git
cd contract-playbook-engine
./start.sh        # macOS/Linux
start.bat         # Windows

This will install dependencies, start the API and frontend, and open your browser. You'll need Python 3.11+ installed.

Option 2: Docker

cp .env.example .env
# Add your ANTHROPIC_API_KEY to .env
docker compose up

Then open http://localhost:8501 in your browser.

Option 3: Hosted Version

Coming soon.

---

What You'll Need

• An Anthropic API key (enter it in the app sidebar or set ANTHROPIC_API_KEY in .env)
• A .docx contract to analyze

The app comes with two sample playbooks pre-loaded (SaaS and Vendor agreements) and sample contracts in data/sample/ so you can test immediately.

---

How It Works

1. Upload a .docx contract
2. Select a negotiation playbook (or upload your own as JSON)
3. Analyze — the tool extracts clauses, compares each against your playbook, and flags deviations
4. Review the results with traffic-light risk ratings (green/yellow/red)
5. Download the deviation report as HTML or Word

Risk Tiers

• Green — Clause aligns with your approved position
• Yellow — Deviation detected, but a fallback position is available. Negotiation recommended.
• Red — Red-line violation or no acceptable fallback. Escalation required.

---

Features

• Ingest structured negotiation playbooks (JSON format) with approved positions, fallback language, red lines, and escalation triggers
• Parse incoming .docx contracts and extract clauses using AI
• Compare each clause against the corresponding playbook position
• Traffic-light risk rating per clause (green/yellow/red)
• Red-line violation and escalation flags
• Suggested fallback language for deviations
• Formatted deviation reports (HTML and Word)
• Multiple playbook profiles (e.g., SaaS, vendor)
• Analysis history with downloadable reports

---

For Developers

Project Structure

contract-playbook-engine/
├── api/                        # FastAPI backend
│   ├── main.py                 # App entry point
│   ├── database.py             # SQLAlchemy setup
│   ├── dependencies.py         # DI (DB sessions)
│   ├── models/                 # SQLAlchemy ORM models
│   │   ├── playbook.py
│   │   ├── contract.py
│   │   └── analysis.py
│   ├── schemas/                # Pydantic schemas
│   │   ├── playbook.py
│   │   ├── contract.py
│   │   ├── analysis.py
│   │   └── report.py
│   ├── routers/                # API routes
│   │   ├── playbooks.py
│   │   ├── contracts.py
│   │   ├── analysis.py
│   │   └── reports.py
│   ├── services/               # Business logic
│   │   ├── document_parser.py
│   │   ├── llm_client.py
│   │   ├── clause_extractor.py
│   │   ├── deviation_analyzer.py
│   │   └── report_generator.py
│   └── templates/
│       └── report.html         # Jinja2 report template
├── cpe_frontend/
│   └── app.py                  # Streamlit UI
├── data/sample/                # Sample playbooks and contracts
├── tests/                      # pytest tests
├── start.sh / start.bat        # One-command launchers
├── Dockerfile                  # API container
├── docker-compose.yml          # Multi-service compose
├── pyproject.toml              # Dependencies
└── README.md

API Documentation

Start the API and visit http://localhost:8000/docs for interactive Swagger documentation.

Key endpoints:

Method	Endpoint	Description
POST	/api/v1/playbooks/	Create a playbook
GET	/api/v1/playbooks/	List playbooks
POST	/api/v1/contracts/upload	Upload a .docx contract
POST	/api/v1/contracts/{id}/extract	Extract clauses (requires API key)
POST	/api/v1/analyses/	Run deviation analysis (requires API key)
GET	/api/v1/reports/{id}/html	Download HTML report
GET	/api/v1/reports/{id}/docx	Download Word report

Running Tests

source .venv/bin/activate
pytest tests/ -v

All LLM calls are mocked in tests — no API key needed to run the test suite.

Tech Stack

• Python 3.11+, FastAPI
• Anthropic API (Claude) for clause analysis
• python-docx for document parsing
• SQLAlchemy 2.0 + SQLite
• Pydantic v2
• Streamlit
• pytest
• Docker + Railway

---

License

MIT