This example demonstrates the complete RUNE pipeline from business requirements to tested code.
REQUIREMENTS.md ──▶ specs/*.rune ──▶ src/*.py ──▶ tests/*.py
(Analyst) (Analyst + AI) (Developer + AI) (Developer + AI)
The analyst produces REQUIREMENTS.md — a plain-language document describing what the system should do. No code, no technical spec. Just business rules and examples.
The analyst feeds the requirements to an AI assistant (Claude, GPT, Cursor) with the RUNE skill loaded. The AI translates each requirement into a formal .rune specification:
Each spec contains the exact signature, behavior rules, constraints, edge cases, and test cases — all derived from the business rules.
The developer gives the .rune specs to their AI coding tool. The AI generates implementations that follow the spec exactly:
From the same specs, the AI generates a complete test suite:
full-project/
├── README.md # This file
├── REQUIREMENTS.md # Step 1: Business requirements
├── specs/
│ ├── calculate_order_total.rune # Step 2: Formal specs
│ ├── validate_coupon.rune
│ └── check_free_shipping.rune
├── src/
│ ├── order_total.py # Step 3: Implementations
│ ├── coupon.py
│ └── shipping.py
└── tests/
├── test_order_total.py # Step 4: Test suites
├── test_coupon.py
└── test_shipping.py
cd examples/full-project
pip install pytest
pytest tests/ -v
The .rune spec acts as a contract between analyst and developer:
The spec is the single source of truth. If requirements change, update the spec, regenerate the code.
specs/src/order_total.pyFor the complete setup guide (what to upload, what prompts to use, how to iterate), see the Workflow Guide.