Tara Framework

File-based spatial applications: develop locally, compile deterministically, and deploy through the existing Tara CLI.

Install

npm install --save-dev @taraspace/framework
npx tframe init my-canvas

npm install
npm run build
npm link
tframe dev examples/knowledge-map

Open http://localhost:4321. Edit a *.node.ts or *.canvas.ts file and the preview reloads.

User workflow

tframe init my-canvas
cd my-canvas
npm install
npm run dev
npm run check
npm run build
tframe deploy <tara-project-id>

For now, tframe deploy is a convenience bridge that compiles .tara/canvas.json, then delegates to:

tara canvas apply <project-id> .tara/canvas.json

The intended product split is tframe dev/build/check for the open framework and tara deploy for the Tara hosting platform, analogous to next versus vercel.

Adding a node

Create it directly:

tframe add node product-idea --type note --text "A new product idea" --color sun

This writes nodes/product-idea.node.ts. A canvas with no nodes array automatically includes every node file. If canvas/main.canvas.ts has an explicit list, add the stable ID:

export default {
  nodes: ["welcome", "product-idea"],
  layout: { type: "grid", columns: 3 },
};

Set TARA_CLI if the tara executable has a different name or path.

Conventions

• nodes/**/*.node.ts: one default-exported Tara node per file.
• canvas/**/*.canvas.ts: canvas composition referencing stable node IDs.
• .tara/: generated manifest, cache, and deployable canvas document.
• Explicit IDs are recommended. When omitted, IDs derive from file paths.

The initial MVP uses built-in Tara Workspace node types. Custom renderer packages and multi-canvas routing are intentionally deferred until the compiler/deployment contract is proven.