Get Started with Rivière

Rivière provides tools for building and querying flow-based architecture graphs.

Schema version: v1.0

Try Éclair First

See what Rivière produces before building your own graph.

Online Demo

View Demo → — Explore a sample e-commerce architecture

Or download the sample: ecommerce-complete.json

Run Locally

git clone https://github.com/NTCoding/living-architecture.git
cd living-architecture
pnpm install
nx serve eclair

Open http://localhost:5173/eclair/ — the demo graph is available at /ecommerce-complete.json

---

What You'll Learn

• How Rivière represents operational flow across your system
• When to use AI extraction, the CLI, or the TypeScript library
• Where to go next for guides, concepts, CLI, and API reference

Why Rivière?

Modern distributed systems are hard to understand. Code is scattered across services, events flow between domains, and tracing an operation from UI to database requires deep system familiarity.

Rivière captures operational flow—not just "what imports what" but "how does placing an order actually work?" From UI click through API, use case, domain logic, events, and handlers.

What is the Output?

Rivière tools produce JSON conforming to the Rivière JSON Schema:

/schema/riviere.schema.json

This schema defines:

• 7 component types: UI, API, UseCase, DomainOp, Event, EventHandler, Custom
• 2 link types: sync, async
• Metadata: domains, sources, entities with state machines

See Resources for example graphs and the visualizer.

Two Paths to Build Graphs

AI Extraction (Recommended)

Use AI + CLI to extract architecture from existing codebases. This is the primary workflow.

Start AI Extraction →

Programmatic Library

Build graphs programmatically in TypeScript. Use when building custom extraction tools or integrating into build pipelines.

Start with the Library →

See Library vs CLI for help choosing.

Why Use the Libraries?

The schema has conventions for IDs, required fields per node type, and validation rules. The libraries:

• Guide you — Type-specific methods surface required and optional fields
• Catch errors early — Invalid configurations fail with helpful messages
• Generate IDs automatically — Following the schema convention
• Validate at build time — Ensuring your graph is well-formed
• Query and analyze — Find components, trace domains, identify entry points

Next Steps

• AI Extraction — Extract architecture from code (recommended)
• CLI Quick Start — Build your first graph with CLI
• Library Quick Start — Build graphs programmatically
• Core Concepts — Components, links, domains explained
• CLI Reference — Command documentation
• API Reference — RiviereBuilder and RiviereQuery documentation

See Also

• Resources
• Library vs CLI