Ask your AI agent to draw — a screen, a flow, a diagram — and it appears in your browser in real time. Edit it, and it follows.
Works with any AI agent you're already familiar with
Claude Code · Codex CLI · Cursor · Gemini CLI · GitHub Copilot · Windsurf
Add FlyingDraw Skills to your project. No installation, no download.
In your desktop app, console, terminal — anywhere you can talk to your agent.
It appears on your canvas instantly. Edit it — your agent keeps up.
A closed feedback loop: your agent draws, you illustrate changes right on the canvas, and your edits flow straight back to your agent — no copy-paste, no screenshots.
One small file teaches your AI tool how to draw. Download it and add it to your tool — no install needed.
→ flyingdraw.com/skill
Sign in with Google to open your own private workspace. This is your board.
Your agent needs two things — your workspace URL and a token. Get both from your avatar menu (top-right of your board):
YOUR-UUID placeholder).In a hurry? Skip the file edit — just paste your URL and token into the chat when your agent prompts you.
Ask your AI agent to "flyingdraw" just about anything — "flyingdraw a login screen" — and watch it appear live on your board.
Using a coding tool? Drop the file in your repo and register it — full per-tool setup on GitHub →
AI agents have gotten remarkably good at thinking — but they can still only talk back in text. The moment an idea turns visual — a screen, a user flow, a system diagram — the conversation breaks down: you paste clunky syntax, copy results into another tool, screenshot it back, and repeat. What makes whiteboarding with a colleague fast — both of you looking at one canvas — is exactly what's missing when you work with an AI agent.
FlyingDraw gives your agent that missing canvas. You say what you want, it draws on a live board in your browser, and you can grab the pen yourself anytime — your agent sees your edits and keeps going. No new app to learn, nothing to export, no copy-paste loop.
| Without FlyingDraw | With FlyingDraw |
|---|---|
| Agent describes a diagram in text or code | Agent draws it on a live canvas |
| You copy it into a separate diagram tool | It appears in your browser instantly |
| Screenshot back and forth to iterate | Edit the canvas — your agent adapts |
| Visual context lost between turns | One shared board you both work on |
Map a campaign funnel, storyboard an email flow, or rough out a landing page.
Wireframe a screen, lay out a user journey, or mock a UI component.
Sketch services, data flows, and deployment diagrams from a prompt.
Generate sequence diagrams, ER schemas, class & state diagrams, and flowcharts — or paste Mermaid or PlantUML and watch it render.
A live whiteboard for AI agents — ask your agent to draw a screen, a flow, or a diagram, and it appears in your browser in real time.
No. Add the skill file to your AI tool, sign in with Google, and ask in plain language — no terminal required.
Yes — it's free to use, and all you need is a Google account.
Wireframes, user flows, flowcharts, org charts, mind maps, and diagrams of all kinds.
Tools like Miro, Lucidchart, FigJam, and Excalidraw are built for drawing by hand, and Mermaid or PlantUML need you to write diagram syntax yourself. Standalone AI generators like napkin.ai or tldraw's Make Real produce a one-off image you then copy elsewhere. FlyingDraw is different: it hands the AI agent you already use — Claude Code, Cursor, Codex, or Gemini CLI — a live canvas it draws on in real time, and any edit you make flows straight back to the agent. No new app to learn, no diagram syntax, and no copy-paste — just ask in plain language and refine together on one shared board.
An AI skill is a small instruction file that gives your AI agent a new ability. The FlyingDraw skill teaches your agent how to turn your words into a drawing and put it on your canvas — you add it to your tool once, with no coding and nothing to install.
A small skill file teaches your agent the drawing format and two API calls to read and write your canvas. It writes the drawing and your browser updates live.
No. FlyingDraw is hosted — you just add the skill file to your tool and connect with a token.
Two layers: Google sign-in in the browser, plus a scoped CLI token (30-day rolling expiry, revocable) that your agent uses. Tokens stay in your chat, never in files.
Yes — switch a workspace to public and anyone signed in with the link can view and edit it together in real time.