pi-tldraw
Supports UIby zeebeecoder
Collaborative canvas app for drawing diagrams and visual sketching integrated with Pi and MCP.
What it does
pi-tldraw packages the tldraw canvas as a Pi extension, allowing agents to open, inspect, edit, and persist visual canvases. It transforms complex drawings into semantic lenses, enabling agents to reason about visual context without processing raw shape data.
Key features
- Semantic Canvas Scenes: Converts drawings into compact graph views for efficient agent reasoning.
- Durable Persistence: Project-scoped snapshots are saved locally in
.pi/tldraw-canvases/. - Direct Agent Control: Tools for opening, executing JS against the editor, and exporting artifacts.
- Visual Export: Export current selections or entire canvases as PNG or SVG for feedback.
Installation
Install via the Pi CLI:
pi install npm:pi-tldraw
Supported hosts
- Claude
- Codex
- Cursor
Quick install
pi install npm:pi-tldrawInformation
- Pricing
- free
- Published
- 7/18/2026
- stars
- 0
Categories
Choose your AI client and follow the steps below.
Codex
pi install npm:pi-tldrawRelated Apps
TomTom MCP Server
MCP AppTomTom's full location stack in your AI client — geocoding, routing, traffic, POI search, and interactive map rendering via MCP.
APS MCP App
MCP AppInteractive 3D design viewer for Autodesk Construction Cloud — browse projects, preview Revit models, and inspect elements via MCP App UI.
Infographic MCP App
MCP AppRenders hand-drawn AntV Infographic diagrams — lists, sequences, timelines, and architecture visuals — directly inside your AI chat.
Chuk MCP UI
MCP App68 standalone MCP UI views — charts, maps, tables, forms and more — each a self-contained HTML app communicating with LLMs via the ext-apps protocol.
Interactive Drawer
MCP AppExcalidraw MCP app — lets any LLM create interactive diagrams with a live browser viewer your users can edit and export.
Visuals MCP
MCP AppRender rich interactive data visualisations inside VS Code Copilot — sortable tables, image galleries, master-detail views, tree views, and drag-and-drop lists.
Nabab
MCP AppInteractive Bayesian network viewer with exact inference, probability sliders, evidence toggling, and MCP server for LLM integration.
ChemCP
MCP AppInteractive 2D molecular structure viewer from SMILES notation, powered by RDKit.js inside Claude.