EventModelers.ai

The Cratis Stack

From a modeled slice to running Cratis code

EventModelers.ai is a visual event modeling platform where business, engineering, and AI agents design a system on one shared timeline — and mark a slice Planned to have an AI coding agent implement it. The Cratis Build Kits are the language-specific agents that pick those slices up and turn them into real Cratis (Arc + Chronicle) vertical slices, built, tested, and committed.

What EventModelers.ai is

Requirements written as tickets lose the reasoning behind them the moment they’re written down, and an AI coding agent handed a vague ticket guesses — confidently, and often wrong. EventModelers.ai closes that gap by making the event model itself the shared artifact: a visual timeline of screens, commands, events, and read models that business and engineering read the same way, and that’s precise enough for an AI agent to build from directly.

The platform’s own loop:

1. Discover & model — map commands, events, and read models onto a shared timeline, in a workshop or an AI-assisted session.
2. Slice the work — break the model into vertical slices, each a self-contained, buildable unit of business value.
3. Generate & build — mark a slice Planned on the board; an AI coding agent implements it, runs the tests, and commits.
4. Stay in sync — change the model and the code catches up, so the board and the codebase never drift apart.

This is the same vocabulary Event Modeling already uses in the rest of the Cratis documentation — commands, events, read models, processors — because EventModelers.ai’s slice types map directly onto Cratis’s own vertical-slice patterns (State Change, State View, Automation, Translation).

Not a Cratis product

EventModelers.ai is built by Nebulit, independent of Cratis. The Build Kits below are what Cratis maintains: the language-specific agents that consume an EventModelers.ai board and generate Cratis code from it. See eventmodelers.ai for the platform itself — workshops, pricing, and the modeling UI live there, not here.

What a Build Kit does

Each Build Kit is a real-time Claude agent you install into a Cratis project. Once connected to a board, it runs a loop that:

1. Connects to the Eventmodelers Platform in real time
2. Fetches a slice’s definition the moment it’s marked Planned
3. Routes it to the right build skill by slice type — a command/event pair, a projection/read model, or a reactor
4. Generates the vertical slice following Cratis conventions for that language
5. Builds and tests it, then updates the board to Done

All without you opening an editor. Each kit’s The Loop and Triggers page has the exact mechanics: C#, Kotlin, and Java. Event Modeling explains why the four slice types map onto Cratis so directly.

Pick your language

Three Build Kits exist today, one per Cratis client stack. Each is a separate, independently versioned repository — pick the one that matches your project:

C#

For .NET projects on Cratis Arc + Chronicle.

Kotlin

For Kotlin projects on Arc.Kotlin + Chronicle.Kotlin.

Java

For Java projects on Arc.Kotlin + Chronicle.Kotlin.

For the C# kit, you need Node.js 18 or later, Claude Code (or its documented Ollama runner), and an EventModelers.ai account with a board. Its installation follows these steps. For Kotlin or Java, follow the language-specific guide linked above; prerequisites and installer behavior may differ.

1. Install in a new, empty directory: npx github:Cratis/Eventmodelers-Build-Kit-CSharp install. It copies a Cratis starter app into the project root and the agent loop into .build-kit/.
2. Connect a board by giving the kit your EventModelers.ai organization, board, and API token (from app.eventmodelers.ai/account).
3. Run the loop with node .build-kit/ralph-claude.js, then mark a slice Planned on your board and watch it land as tested code.

Before you install the C# kit

The C# installer writes the starter app’s files (such as CLAUDE.md, package.json, and the project and solution files) into the current directory and overwrites files with the same names. Run it in a new directory, not in an existing project.

Your API token is saved in .eventmodelers/config.json in the project root. The installer adds .build-kit/ to .gitignore, but not that file. Add .eventmodelers/ to .gitignore yourself before you commit, so the token does not end up in your repository.

See also

• Event Modeling — the notation and the four slice patterns the kits generate from
• eventmodelers.ai — the platform: workshops, the modeling UI, and the FAQ
• Build a full-stack feature — the same command/event/read-model shapes, built by hand