How Our Tools Work Together
Synapify
Synapify
The visual editor for RIDDL. Everything inside this box is
Synapify: start a model from a brief or a template, author and
refine it visually or as text, and simulate it — all
against continuous validation. It edits the
.riddl model directly, so the
file stays the artifact.
Requirements
Define Requirements
Describe your domain in plain English and let AI suggest an initial RIDDL structure — domains, contexts, entities, commands, events, and workflows. Start from what the system does, not technology choices.
The drafting is done by riddlg, which turns the brief into validated RIDDL.
Learn more →Template
Choose a Template
Pick an industry-specific starting point from riddl-models — healthcare, finance, commerce. Customize proven patterns rather than starting from scratch.
Learn more →Refine Validates Continuously
Author & Refine
Visual + text editing with bidirectional sync. Drag and drop domains, contexts, entities, and workflows, or write RIDDL directly. Validation runs continuously using the same engine as the riddlc compiler.
AI assists with pattern suggestions, error explanations, and completeness checks.
Learn more →Simulate & Revise Cycle
Run your model through the simulator to validate behavior, not just structure. When issues are found, revise the model in Author & Refine, then simulate again — without leaving Synapify.
Simulator
Runs your model — exercising entities, message flows, and state machines without a line of implementation code. Validates behavior, not just structure.
From 1.0.0 this is built into Synapify, not a separate tool.
Learn more ↓
riddlg
riddlg
The local-first generator, and the engine behind both directions of the arrow to its left. It turns a natural-language brief into validated RIDDL, and turns a validated model into documentation, API specifications, and code. Everything runs on your machine.
Three ways in: the command line, an MCP server your AI assistant drives, and a local HTTP service.
About riddlg →Command line
riddlg gen docs,
gen api,
gen code and friends —
scriptable, and the same binary your CI runs.
MCP server
riddlg mcp exposes validation,
analysis, and generation as tools your AI assistant can call
directly — Claude, Cursor, or anything else speaking MCP.
HTTP server
riddlg serve runs a local HTTP
and WebSocket service, so Synapify and other tools can drive
generation over an API without shelling out.
Generate Documentation
Always-in-sync documentation from your specification — architecture overviews, entity catalogs, event flows, and diagrams. Never maintain separate docs that drift from reality.
Learn more →Generate API Specifications
OpenAPI, gRPC, and Smithy specifications generated straight from your contexts — so the contract your consumers integrate against is derived from the model, not hand-written beside it.
Learn more ↓Generate Code
Entity classes, command handlers, event definitions, API endpoints, and test stubs from the specification. Developers implement business logic in the placeholders, then compile, test, and deploy with their existing toolchain.
Learn more ↓
riddlc
riddlc
The open source RIDDL compiler: it parses and validates the
.riddl model, and produces the AST
everything else works from. Synapify embeds it for continuous
validation and riddlg reads models through it — so a model
that is valid in one is valid in all of them.
Free and open source, and usable on its own from the command line.
About riddlc → .riddl model — one source of truth
riddlc parses and validates it — the same engine inside both tools
Start from
orGenerates
.riddl model — one source of truth
riddlc parses and validates it — the same engine inside both tools
From business idea to deployed system — Synapify, riddlc, and riddlg work on one shared model, so your design stays the source of truth.
Why This Approach Matters
Shared Understanding
Business experts and developers work with the same artifact. No translation layer between “what we designed” and “what we built.”
Risk Reduction
Validate structure and simulate behavior before writing implementation code. Find design flaws when they cost minutes to fix, not weeks.
Living Documentation
The RIDDL specification is the source of truth from day one through production. It evolves with the system instead of decaying into outdated artifacts.
AI-Native Design
RIDDL models are structured data, not diagrams or prose. AI tools can read, generate, and reason about RIDDL directly — making your architecture a first-class input to code generation, validation, and simulation rather than a picture someone has to interpret.
Local Inference
Running a model locally is riddlg’s default, not its fallback: your domain description never has to leave your machine, and generation costs only the electricity to run it — no per-token API bills. Gemma ships as the default because it produces the best RIDDL in our testing; point riddlg at a hosted model instead whenever you prefer. Apple Silicon suits local work especially well — CPU and GPU share one pool of memory, so models too large for a discrete card’s VRAM still run on a laptop.
Incremental Adoption
You don’t have to model your entire enterprise on day one. Start with a single bounded context, prove value, and expand. Each RIDDL model is self-contained and composable with others as your practice matures.
Compliance-Ready
RIDDL’s built-in briefly and described by
clauses, along with term definitions and author
metadata, produce models that double as living specification documents.
Auditors and stakeholders can read the model directly — no
separate documentation to keep in sync.
Onboarding Accelerator
New team members use Synapify to explore the system’s domains, contexts, and message flows through graphical views and guided navigation — grasping the architecture in hours instead of weeks. The model is the single source of truth, not a wiki, not tribal knowledge, not a stale diagram.
Technology Independence
RIDDL describes what your system does, not how it’s built. Swap databases, messaging platforms, or cloud providers without rewriting your architecture. The model stays stable while implementations evolve.
Simulation: Your Model’s Digital Twin
Synapify creates a running digital twin of your RIDDL model — exercising every entity, message flow, and state machine without a single line of implementation code.
(the spec)
(simulation)
Steer
(Synapify)
Five Simulation Layers
Each layer is opt-in with sensible defaults. Start with system model validation and add layers as your design matures.
System Model
Exercises entity state machines, command handlers, and message flows exactly as your RIDDL model defines them. Auto-generates tests from the model.
Timing
Models realistic latency — fast intra-context messaging versus slower cross-context network hops. Configurable profiles reveal timing-sensitive design issues.
Chaos
Injects failures — dropped messages, timeouts, partial outages — to test how your design handles the unexpected before production does it for you.
Infrastructure
Models hardware as general classes — compute, storage, networking — to validate capacity and persistence assumptions at the architecture level.
Cost
Estimates cloud resource consumption based on simulated load. Catch expensive design decisions at the whiteboard stage, not the invoice stage.
Epics Drive Scenarios
User journeys already in your RIDDL model become simulation scenarios. Define how many, when, and what mix — don’t re-describe the flows.
AI-Authored Scenarios
AI reads your model and your intentions to generate realistic test scenarios — including edge cases you might not think to write by hand.
No Code Required
The RIDDL model is the spec, the digital twin is the simulation, Synapify is the control plane. Validate your design before writing implementation code.
Generation: From Model to Artifacts
One validated model, twenty artifacts. riddlg generates every one of these from the same source of truth, so they cannot drift from each other or from the design. Synapify drives it for you; you can also run any of it yourself from the command line.
Documentation
riddlg gen docs AsciiDoc
AvailableThe default. AsciiDoc sources with domain maps, entity catalogs, and message-flow diagrams.
MkDocs
AvailableA Material-ready MkDocs site, so the model publishes straight into your existing docs pipeline.
Hugo Book
AvailableA Hugo site on the Book theme — navigable documentation from the model alone.
Hugo Geekdoc
AvailableThe Geekdoc variant, and what plain hugo resolves to.
DocBook
ProDocBook XML, for toolchains that publish to PDF, print, or a formal documentation system.
DITA
ProDITA topics, for structured-authoring environments that expect them.
API Specifications
riddlg gen api Smithy
AvailableThe default. Smithy models derived from your contexts, ready for AWS codegen.
OpenAPI
AvailableOpenAPI descriptions of the operations each context exposes.
gRPC
AvailableProtocol Buffers service definitions, generated from the same contexts.
AsyncAPI
AvailableEvent-driven contracts — Kafka, AMQP, MQTT, NATS or WebSocket, per context.
JSON Schema
AvailableJSON Schema for the types in your model, for validation anywhere JSON travels.
Data & Schema
riddlg gen sql · gen dbml SQL DDL
AvailableOne .sql file per entity — PostgreSQL, MySQL, Oracle, SQL Server or ANSI.
DBML
AvailableA logical schema you can paste straight into dbdiagram.io.
Catalogs & Portals
riddlg gen backstage · gen catalog · gen confluence Backstage
AvailableA catalog-info.yaml Software Catalog, with ownership taken from the model.
EventCatalog
AvailableAn EventCatalog site documenting your domains, services and the messages between them.
Confluence
ProA Confluence storage-format export, with a publish script for your space.
Model
riddlg gen riddl RIDDL
AvailableThe other direction: a natural-language brief becomes validated RIDDL, drafted locally.
Code
riddlg gen code Quarkus / Java
ProRunnable Quarkus sources with AI-filled, compile-verified handler bodies.
Scala / Pekko
Coming SoonThe same pipeline, targeting Scala and Pekko actors.
TypeScript / Effect
Coming SoonThe same pipeline, targeting TypeScript with Effect.
See It in Action
Try the RIDDL language in your browser, explore the documentation, or see how it fits your team's needs.