This is an elevator simulation built with Elixir and Phoenix LiveView. I've built this to learn Elixir, and got to exercise the following:
- the Actor Model
- a pure functional architecture
- Augmented AI developmen with Zed, and Sonnet 4.6
- Multiagent orchestration with Antigravity and Gemini 3 Flash
- BDD with gherkin features and scenarios
- TDD with regular Exunit
- End-to-end (e2e) testing with Playwright and ExUnit
- Continuos integration (CI/CD) with Github actions.
- Quality gates with static and security analysis (OWASP)
- Deployment with Docker on fly.io.
All specs are inside doc/specs/
- Rules: Business logic in
rules.md. - Behaviors: Observable behavior is defined in formal Gherkin feature files within the
features/directory. - Traceability: Every test is explicitly linked to a Scenario ID (e.g.,
[S-MOVE-WAKEUP]). No code is written until a failing test proves the need for it. Seetraceability.md
The system is built around a brain containing all the logic and simple and quite dumb components collaborating through a message bus. See architecture.md
- Elixir (OTP): Using lightweight processes and immutable state for concurrent fault-tolerant logic.
- Phoenix LiveView: A real-time, "No-Build" frontend architecture. With vanilla CSS and standard JavaScript.
- Mise: Automated toolchain management to ensure all developers (and agents) use the exact same versions of Elixir, Erlang, and Node.js.
- ExUnit: logic proofs for the Brain and parts of the Servo.
- Gherkin features: Steps imlemented with Cabbage into plain ExUnit.
- Integration testing: A couple of tests testing all components in a fast simulation.
- Playwright: "Happy Path" in a real browser environment.
See an overview of all available documentation.
Created by Alex Schenkman, on April 2026, and assisted by Gemini 3 and Sonnet 4.6