tldraw-offline-mcp
Enables remote agents to control tldraw Desktop's Canvas API, manage documents, scripts, assets, and take screenshots.
README
tldraw-offline MCP
A Bun + Hono MCP server that gives a remote agent complete control of tldraw Desktop's local Canvas API, durable document-script workspaces, assets, and screenshots.
The server runs on the Mac beside tldraw. Remote clients receive virtual workspace paths and inline MCP images—never unusable Mac paths or the app's bearer token.
TODO
- should be managed by
neo - efficiency around mcp docs reading requests
Requirements
- macOS with tldraw Desktop and its Canvas API
- Bun
- an MCP client supporting Streamable HTTP
- Tailscale for the recommended remote transport
Install
cd /Users/c1re/Developer/infra/mbp/tldraw-offline-mcp
bun install
bun run check
Start
Export a stable bearer token, then run the start script:
export TLDRAW_MCP_TOKEN="$(bun run --silent token)"
bun run start
The script starts the MCP server on 127.0.0.1:7237 and runs Tailscale Serve in HTTP-only mode on port 80. It prints the remote MCP URL and keeps both processes in the foreground until you press Ctrl-C.
Local endpoints:
http://127.0.0.1:7237/mcphttp://127.0.0.1:7237/healthz- authenticated
http://127.0.0.1:7237/readyz
Test the local endpoint with the official MCP client:
TLDRAW_MCP_URL=http://127.0.0.1:7237/mcp \
TLDRAW_MCP_TOKEN="$TLDRAW_MCP_TOKEN" \
bun scripts/smoke.ts
Configure a remote MCP client
Copy the printed HTTP endpoint and bearer token to the trusted Linux host. A generic Streamable HTTP configuration looks like:
{
"mcpServers": {
"tldraw-offline": {
"url": "http://DEVICE.TAILNET.ts.net/mcp",
"headers": {
"Authorization": "Bearer YOUR_MCP_TOKEN"
}
}
}
}
The exact configuration location depends on the MCP host. Keep the endpoint restricted with Tailscale ACLs as well as the bearer token.
Install skill/tldraw-offline/SKILL.md on the Linux agent host. If this repository is present there:
./scripts/install-skill.sh
Tool surface
Discovery and reference
tldraw_healthtldraw_docs_listtldraw_doc_focusedtldraw_doc_shapestldraw_doc_bindingstldraw_reference_searchtldraw_imports_searchtldraw_helpers_listtldraw_recipes_listtldraw_recipe_gettldraw_readme
Canvas control
tldraw_search— complete Canvas API escape hatchtldraw_exec— complete live Editor escape hatchtldraw_linttldraw_screenshot— inline JPEG MCP content
Durable scripts and assets
tldraw_workspace_opentldraw_workspace_listtldraw_workspace_readtldraw_workspace_applytldraw_script_statustldraw_script_error_log
The workspace API exposes only virtual script/** and assets/** paths for writes. Existing files use SHA-256 preconditions, exact edits require one unique match, complete buffers are written in watcher-compatible order, and entrypoints are committed last.
Configuration
Copy .env.example to .env, set TLDRAW_MCP_TOKEN, and update any values you want to override.
Development
bun run dev
bun run typecheck
bun test
bun run check
operator-guide/guide.md is served as tldraw://guide. The MCP-native agent skill is under skill/tldraw-offline/.
Cool Examples
- https://x.com/max__drake/status/2078151243650810198?s=20
- https://x.com/tldraw/status/2078164903911719305?s=20
- https://x.com/shubgaur/status/2078309824199074020?s=20
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.