Ithaca

Interactive AI teaching platform for visualizing algorithms, pseudocode, and step-by-step reasoning.

Ithaca is a classroom-oriented teaching tool for university AI courses. Its central principle:

Show the reasoning process, not only the final answer.

Instead of just displaying a final path or answer, Ithaca makes every intermediate decision visible — the current frontier, why a node was chosen, how the pseudocode maps to execution, and how the answer was actually reached.

Live demo →

Features

• Step-by-step execution — Step Forward / Step Backward / Run / Pause / Reset, all backed by precomputed, deterministic execution snapshots so backward stepping is exact.
• Synchronized views — the graph, state panel, pseudocode (active-line highlighted), step explanation, and decision trace all update together for every step.
• Presenter-first controls — Page Up / Page Down drive stepping directly, so a lecturer can run the whole class from a presentation remote. Full keyboard shortcut list via ?.
• Two graph examples:
  • Weighted Graph — multiple paths with different costs and heuristics, tuned so BFS, DFS, UCS, Greedy, and A* each find a genuinely different path.
  • Tree — a single path to every node, isolating the difference in traversal order between BFS (level-by-level) and DFS (depth-first, with backtracking).
• Light / dark theme, fullscreen presentation mode, and a layout designed for classroom projectors.

Algorithms

Algorithm	Frontier	Notes
Breadth-First Search	FIFO queue	Not cost-sensitive — finds the fewest-edges path
Depth-First Search	LIFO stack	Not generally optimal
Uniform-Cost Search	Priority queue by g(n)	Optimal by path cost
Greedy Best-First Search	Priority queue by h(n)	Ignores path cost so far — can be misled
A* Search	Priority queue by f(n) = g(n) + h(n)	Optimal given an admissible heuristic

All five share one execution engine and one graph visualization, so a lecturer can reset and switch algorithms to compare behavior directly on the same example.

Tech stack

• Svelte 5 + TypeScript + Vite
• Tailwind CSS v4
• shadcn-svelte-style primitives on bits-ui
• Svelte Flow (@xyflow/svelte) for graph visualization
• Lucide icons
• Vitest for algorithm correctness tests

Ithaca is a frontend-only static app — no backend, no server-side rendering.

Getting started

npm install
npm run dev

Other scripts:

npm run build   # production build to dist/
npm run preview # preview the production build locally
npm run check   # svelte-check + TypeScript project check
npm test        # run the algorithm test suite (Vitest)

Project structure

src/
├── lib/
│   ├── algorithms/search/   # BFS, DFS, UCS, Greedy, A* — pure functions, UI-independent
│   ├── execution/           # Step controller: forward/backward/run/pause/reset
│   ├── graph/               # Graph types + the two teaching examples
│   ├── shortcuts/           # Centralized keyboard shortcut dispatch
│   └── theme/                # Light/dark theme store
└── components/
    ├── layout/              # App shell, header, sidebar, theme/fullscreen toggles
    ├── visualization/       # Svelte Flow graph, custom node/edge components
    ├── state/               # Queue/Stack/PriorityQueue views
    ├── pseudocode/          # Active-line pseudocode panel
    ├── trace/               # Step explanation + decision trace
    ├── execution/           # Playback controls, shortcuts help dialog
    └── ui/                  # Reusable primitives (Button, Dialog, Tooltip, ...)

Algorithm logic is fully decoupled from the UI and tested independently — see src/lib/algorithms/search/__tests__.

Keyboard shortcuts

Key	Action
Page Down / →	Step forward
Page Up / ←	Step backward
Space	Run / Pause
R	Reset
?	Show shortcuts
Esc	Close dialog / exit fullscreen

Deployment

Pushing to main builds and deploys the app to GitHub Pages automatically via .github/workflows/deploy.yml.

Roadmap

Search algorithms are the first of several planned AI topics: Genetic Algorithm, Propositional Logic, First-Order Logic, Prolog, K-Nearest Neighbor, and Naïve Bayes — sharing the same execution and visualization architecture.

License

MIT — see LICENSE.