---
url: https://chartbuddy.io/hub/docs/guides/desktop-basics.md
---
> Docs index: [llms.txt](https://chartbuddy.io/hub/docs/llms.txt)

# Desktop basics

Chartbuddy Hub is your local chart editor. You operate it from your AI chat — Hub itself has no prompt box. MCP is the remote control; Hub is where charts live and get edited.

## First launch

1. Install Hub to **Applications** from the `.dmg`
2. Open Chartbuddy once (approve Gatekeeper if asked)
3. Confirm the bridge is up: the Hub `mcp-port` file should contain a port (macOS / Linux `~/.chartbuddy/mcp-port`, Windows `%USERPROFILE%\.chartbuddy\mcp-port`)
4. Connect your AI client — see [Connect MCP](https://chartbuddy.io/hub/docs/connect-mcp.md)

Leave Hub installed under `/Applications` so MCP auto-launch can find it.

## Typical workflow

1. **Prompt in AI** — ask Cursor, Claude, ChatGPT, or another MCP-connected client to create or update a chart
2. **Refine in Hub** — focus the chart window for pixel-level control (toolbar, spreadsheet panel, context menus)
3. **Push out** — send the live chart to Google Slides, Notion, Slack, or export as PNG
4. **Pull back** — grab a Chartbuddy PNG from another surface and pull it into Hub to edit again

## What Hub shares with the Chrome extension

Hub packs the same chart engine as the Chartbuddy for Google Slides™ extension:

* Waterfalls, mekkos, stacked bars, CAGRs, and the full business library
* Native editing — not a static screenshot
* Brand styling and number formats

The desktop shell adds push/pull, MCP integration, and a standalone window without a browser tab.

## Next

* [Connect MCP](https://chartbuddy.io/hub/docs/connect-mcp.md) — Cursor, ChatGPT, Claude, …
* [Push & pull](https://chartbuddy.io/hub/docs/guides/push-and-pull.md) — moving charts between surfaces