Sample GenLayer project

About

This project includes the boilerplate code for a GenLayer use case implementation, specifically a football bets game.

What's included

• An example intelligent contract (Football Bets) with web access and LLM integration
• Direct mode tests — fast, in-memory unit tests with web/LLM mocking (~ms per test)
• Integration tests — full end-to-end tests against GenLayer Studio
• Contract linting — static analysis to catch common contract issues before deployment
• CI pipeline — GitHub Actions workflow for linting and direct tests
• A production-ready Next.js 15 frontend with TypeScript, TanStack Query, and Radix UI
• Configuration file template and deployment scripts

Requirements

• Python >= 3.12
• GenLayer CLI globally installed: npm install -g genlayer
• GenLayer Studio (for integration tests and deployment): Install from Docs or use the hosted GenLayer Studio

Project Structure

contracts/              # Python intelligent contracts
tests/
  direct/               # Fast in-memory tests (no Studio required)
    test_create_bet.py   # Bet creation logic
    test_resolve_bet.py  # Bet resolution with web/LLM mocks
    test_views.py        # Read-only view methods
  integration/           # Full tests against GenLayer Studio
    test_football_bets.py
    fixtures.py          # Expected state fixtures
frontend/               # Next.js 15 app (TypeScript, TanStack Query, Radix UI)
deploy/                 # TypeScript deployment scripts
gltest.config.yaml      # Test runner network configuration
pyproject.toml          # Python/pytest configuration
.github/workflows/      # CI pipeline

Quick Start

1. Set up Python environment

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

2. Lint your contracts

Run the GenVM linter to catch issues before deployment:

genvm-lint check contracts/football_bets.py

The linter catches:

• Forbidden imports and non-deterministic calls
• Invalid storage types (must use TreeMap, DynArray, u256, etc.)
• Missing decorators and return type annotations
• Non-deterministic operations outside equivalence principle blocks
• And 20+ other rules

3. Run direct mode tests

Direct mode tests run contracts in-memory without needing GenLayer Studio. They use mocks for web requests and LLM calls, giving you fast feedback (~milliseconds per test):

pytest tests/direct/ -v

Direct mode features used in these tests:

• direct_deploy("contracts/file.py") — deploy contract in memory
• direct_vm.sender = address — set transaction sender
• direct_vm.mock_web(pattern, response) — mock HTTP/render calls
• direct_vm.mock_llm(pattern, response) — mock LLM responses
• direct_vm.expect_revert("message") — assert expected failures
• direct_vm.clear_mocks() — reset mocks between calls

4. Deploy the contract

1. Choose your network: genlayer network
2. Deploy: genlayer deploy (runs the script in /deploy/deployScript.ts)

5. Run integration tests

Integration tests deploy the contract to GenLayer Studio and test with real consensus:

gltest tests/integration/ -v -s

These require GenLayer Studio running (local or hosted).

6. Set up the frontend

1. Copy frontend/.env.example to frontend/.env
2. Add your deployed contract address as NEXT_PUBLIC_CONTRACT_ADDRESS
3. Run:

cd frontend
npm install
npm run dev

The app will be available at http://localhost:3000/.

How the Football Bets Contract Works

1. Creating Bets: Users bet on a football match by providing the game date, teams, and predicted winner.
2. Resolving Bets: After the match, the contract fetches results from BBC Sport, uses an LLM to extract the score, and validates via the equivalence principle.
3. Points: Correct predictions earn points. Users can query their points or the leaderboard.

Testing Strategy

Test Type	Command	Speed	Requires Studio
Lint	genvm-lint check contracts/*.py	~250ms	No
Direct	pytest tests/direct/ -v	~ms/test	No
Integration	gltest tests/integration/ -v -s	~min/test	Yes

Recommended workflow:

1. Lint after every contract change
2. Run direct tests frequently during development
3. Run integration tests before deployment to verify consensus behavior

For AI coding agents (Claude Code, Cursor, etc.), the linter and direct tests provide the fast feedback loop needed for iterative development without requiring a running Studio instance.

Community

• Discord: Discussions, support, and announcements
• Telegram: Informal chats and quick updates

Documentation

For detailed information, see our documentation.

License

This project is licensed under the MIT License - see the LICENSE file for details.