What problem does it solve? Working with open tldraw Desktop documents programmatically requires knowing the app's local HTTP server, its per-launch bearer token, and which endpoints read shapes, execute editor code, or install durable board scripts. This Skill provides the complete operating procedure so an agent can inspect, edit, connect, lint, and script a live canvas without manual GUI work. ## Core Features & Use Cases - Canvas inspection and editing: Discover open documents, read shapes and bindings, and mutate the canvas via the /api/search and /api/doc/:id/exec endpoints, with screenshots for visual verification. - Durable board scripts: Create persistent behavior (clickable UI, animations, custom shapes, bindings, overlays) through the /script-workspace endpoint and script/main.js / script/config.js files, with watcher status checks. - Safe collaboration handling: Distinguish local, remote, and server document ownership, gate shared-board script writes on host status, and save only locally owned documents. - Use Case: Ask the agent to arrange and connect boxes on your open project-plan canvas, add a clickable card UI that persists across sessions, then verify the result with a screenshot and lint pass. ## Quick Start Ask the agent to list the shapes on your currently open tldraw canvas and connect two of them with a bound arrow.
