Quick Start

The fastest way to evaluate rulesgen is the Docker Compose stack started by ./scripts/run_stack.sh. It starts the API, OpenSandbox support, and an LLM translation path. If no provider credential is present, the script falls back to the built-in stub translation backend so the stack can still run locally.

Prerequisites

• Docker
• docker compose
• curl
• jq
• Optional: one LLM provider credential such as OPENAI_API_KEY, ANTHROPIC_API_KEY, GEMINI_API_KEY, or AZURE_API_KEY

Start the stack

./scripts/run_stack.sh

When no provider key is set, the script sets RULESGEN_LLM_GATEWAY_BACKEND=stub. The stub backend is useful for local smoke tests and API walkthroughs. Set a provider key when you want real natural-language translation.

Service endpoints

• API root: http://127.0.0.1:8000
• Swagger UI: http://127.0.0.1:8000/docs
• ReDoc: http://127.0.0.1:8000/redoc
• Versioned Swagger UI: http://127.0.0.1:8000/v1/docs

Verify readiness

curl -s http://127.0.0.1:8000/health/ready

The readiness response confirms that the API process is accepting requests.

Set a base URL

The workflow examples use BASE_URL for readability.

export BASE_URL=http://127.0.0.1:8000

What to do next

• Follow Example Workflows to parse, compile, preview, generate, poll, and download artifacts.
• Use API Reference when integrating with the HTTP API.
• Use Python Library when embedding rulesgen directly in another Python process.
• Review Configuration, Run Modes, and Safety Guardrails before operating the service beyond a local evaluation.
• Use Repository Docs for the longer design and contributor material that still lives in the repository.