Skip to content

VISION.md

Latest commit

 

History

History
82 lines (60 loc) · 2.59 KB

VISION.md

File metadata and controls

82 lines (60 loc) · 2.59 KB

Project Vision

What is this?

GroupMixer is a web application for distributing people into groups based on constraints and preferences. It has a Rust backend (gm-core, gm-api, gm-wasm) and TypeScript frontend (webapp).

Goals

Primary: 100% AI-Testable Application

Make the entire application end-to-end testable by AI agents without human intervention. AI should be able to:

  1. Visual Testing

    • Render every screen/route of the app
    • Render all modals and overlays
    • Test on all screen sizes (mobile, tablet, desktop)
    • Take screenshots and visually verify layout/styling
    • Detect visual regressions automatically

  2. Functional Testing

    • Test all user interactions (clicks, inputs, drag-drop)
    • Verify all API endpoints work correctly
    • Test error states and edge cases
    • Validate data flow end-to-end

  3. Full Stack Coverage

    • Frontend: All UI components, routing, state management
    • Backend: All API endpoints, business logic, solver algorithms
    • Integration: Frontend ↔ Backend communication

Secondary Goals

  • Self-documenting test coverage
  • Reproducible test environments
  • Fast feedback loops for AI debugging

Current State

  • Rust backend with gm-core, gm-api, gm-wasm
  • TypeScript webapp frontend
  • Existing CLAUDE.md for coding conventions
  • No comprehensive AI-friendly testing infrastructure yet

Roadmap

Near-term (Next Features)

  • Research existing AI E2E testing approaches for similar stacks
  • Set up screenshot-based visual regression testing
  • Create programmatic access to all UI states
  • Implement API contract testing

Long-term

  • Full autonomous debugging by AI agents
  • AI can identify, diagnose, and fix issues without human help
  • Continuous AI-driven quality assurance

Technical Constraints

  • Stack: Rust backend, TypeScript/JS frontend
  • Must work in CI/CD pipelines
  • Should support headless browser automation
  • Must be deterministic and reproducible

Research Areas

  • Playwright/Puppeteer for browser automation
  • Visual regression testing tools (Percy, Chromatic, etc.)
  • Storybook for component isolation
  • API testing frameworks (for Rust backend)
  • AI-specific testing patterns used in similar projects
  • How other Rust+TS webapps achieve full test coverage

Feature Guidelines

  • Features should enable AI agents to test autonomously
  • Prefer programmatic interfaces over manual testing
  • All UI states should be reachable via URL or API
  • Tests should be self-verifying with clear pass/fail criteria

Not Now

  • Performance optimization (focus on testability first)
  • New user-facing features (infrastructure first)