VS Code Extension  ·  Beta

Your diagrams, as code.

A VS Code canvas editor that saves diagrams as AI-friendly JSON.

Edit visually, commit cleanly, collaborate with AI naturally.

Install for VS Code
architecture.jis.json — Jiscribe 
{
  "$schema": "schema.jiscribe.dev/...",
  "root": [
    { "id": "server", "type": "rect",
      "text": "API Server",
      "x": 100, "y": 80, "width": 160, "height": 80 },
  ],
  "connectors": [{ "id": "c1", "type": "connector", ... }]
}

Explaining architecture to AI is painful.

WHY JISCRIBE

Jiscribe fixes this.

AI

AI-Friendly Format

Schema-backed JSON that LLMs read and write natively. No PNG exports or SVG tangles — just clean, semantic data every AI understands.

SYNC

Real-time Two-Way Sync

Drag a shape on the canvas — JSON updates instantly. Let AI autocomplete the JSON — canvas renders in real-time. Both directions, always in sync.

GIT

Git-Friendly Diffs

Diagrams live in your repo as neatly formatted JSON. Review "which shape changed" in a PR just like any other code file.

VSC

Everything in VS Code

No context-switching to Figma or Draw.io. Draw system diagrams in a tab right next to your source code.

Get started in seconds.

1

Install Extension

Find "Jiscribe" in VS Code Marketplace and install.

2

Create a .jis.json File

Right-click in Explorer › New Empty Jiscribe Canvas.

3

Draw or Prompt AI

Use the canvas UI directly, or let AI write the JSON for you.

A format built for LLMs.

Clean, semantic JSON any LLM can generate and parse. Every shape has meaning — not just pixels.

schema.jiscribe.dev/v1/jiscribe.schema.json
architecture.jis.json
{
  "root": [
    { "id": "server", "type": "rect",
      "text": "API Server",
      "x": 100, "y": 80, "width": 160, "height": 80 }
  ],
  "connectors": [
    { "id": "c1", "type": "connector",
      "source": {...}, "target": {...} }
  ]
}