It runs discrete-event experiments on individual requests rather than drawing plausible-looking diagrams, so latency, queues, and failures are measured. Build a topology on a canvas, raise traffic or inject faults, and inspect where capacity collapses. It also ships as a VS Code panel and an MCP server for AI assistants.
For System design learners building intuition about failure cascades, and developers integrating AI architecture analysis
A pure discrete-event engine models individual requests moving through a topology, while React only presents its measured snapshots. Finite slots, FIFO queues, variable service times, and abandoned downstream work make overload and retry cascades emerge from simulated events.
README.md (How the simulation works); AGENTS.md; src/sim/annotations.ts
Simulation engine
Simulates topology traffic without React, DOM, or I/O.
src/sim/
Preset library
Defines teaching scenarios and reconstructed architecture topologies.
src/sim/presets.ts
Interactive canvas
Draws and edits nodes, links, selections, and annotations.
src/components/Canvas.tsx
Application shell
Coordinates layout, persistence, history, sharing, and simulation presentation.
src/App.tsx
Generated
