@fusedashlabs/ui9000-workspace
OfficialSupports spatial intents by producing Mapbox-powered map charts, with a configurable Mapbox access token for rendering maps.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@@fusedashlabs/ui9000-workspaceshow a comparison chart for the attached CSV"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
@fusedashlabs/ui9000-workspace
MCP server with one tool, show_workspace. Give it a table and a question. It returns a chart the host can draw. The cells stay on the server. The result is a chart spec plus a signed data link, not the rows.
npm: @fusedashlabs/ui9000-workspace
What it does
show_workspace reads one table, classifies the columns, and draws one chart.
When
TYPESAFE_API_KEYis set, Jev chooses a class from the columns, then a chart inside that class. The result says why.A named chart in that class is the one drawn.
A named chart in another class returns
awaitingUser. Nothing is drawn.suggestionis the chart Jev would draw. The next call passes that chart asrequestedChartwith the samedatasetId.A named chart this server cannot draw returns
awaitingUserand does not draw a substitute.If Jev is unset, unreachable, or does not choose, the server picks a chart from the table.
The tool call must not include the table cells as data, rows, points, or series.
Related MCP server: Vela MCP Server
Connect
Node.js 20 or newer for the npx install. Jev is optional. Without TYPESAFE_API_KEY the server still draws a chart. Put the key only in the env block below. Do not commit it.
On connect the server returns instructions for Cursor, Claude Desktop, and Claude.ai. The host puts that text in context before it picks a tool: for a table or a chart, call show_workspace and do not draw with the chat's built-in chart tools. Those other tools stay available. A host that ignores server instructions can still use them.
Cursor
~/.cursor/mcp.json:
{
"mcpServers": {
"UI9000-Workspace": {
"command": "npx",
"args": ["-y", "@fusedashlabs/ui9000-workspace"],
"env": {
"TYPESAFE_API_KEY": "your-typesafe-key"
}
}
}
}Restart Cursor after saving.
Claude Desktop
claude_desktop_config.json:
{
"mcpServers": {
"UI9000-Workspace": {
"command": "npx",
"args": ["-y", "@fusedashlabs/ui9000-workspace"],
"env": {
"TYPESAFE_API_KEY": "your-typesafe-key"
}
}
}
}Claude.ai
Claude on the web cannot run npx. Add a custom connector:
Customize → Connectors → Add custom connector
Name:
UI9000-WorkspaceURL:
https://mcp.ui9000.com/workspace/mcp
Enable the connector in the chat. This hosted server accepts csv or datasetId. It does not accept url or path. The Jev key is not part of this connector. Use the npx install above when you want to pass TYPESAFE_API_KEY.
How to call show_workspace
intent is required. It is the question, not a chart name.
Intent | Use it for |
| Where something is: region, country, city, lat/lng. A map needs geo columns. |
| Which group is higher or lower, or how values spread. |
| A total, a count, or a few headline numbers. |
| The user must enter or confirm labelled values. |
| Claims, sources, or a timeline of facts. |
| What is connected to what. |
Pass one source:
Field | When |
| The table is in the message, header included. |
| A local or dev file URL. Not accepted on the hosted connector. |
| A local file path. Not accepted on the hosted connector. |
| The same table as an earlier call. |
Optional:
Field | When |
| The user's words, so Jev can read them. |
| Only when the user named a chart, such as |
| Header names to keep. Later calls on the same |
Keep datasetId from the result for the next chart on the same table.
Environment
Variable | Default | What |
| unset | When set, Jev chooses the chart if the user named none. |
| shipped demo table | Table used only when the call has no |
| a public map token | Lets a map draw. Set another token to replace it. |
|
| Where chart data is stored. |
|
| How long a data link stays valid. Minimum 1. |
|
| Largest table the server will store. |
This server cannot be deployed
Maintenance
Related MCP Connectors
- BasedashOAuthcom.basedash
Governed BI MCP. Ask questions of live company data and list workspace sources via OAuth.
Renders interactive Chart.js charts and dashboards inline in AI conversations.
Create, inspect, manage, and render charts and data visualizations as SVG/PNG or interactive embeds.
Assess AI-discovery readiness, plan visibility fixes, and summarize scan evidence. Read-only.
Related MCP Servers
- FlicenseBqualityDmaintenanceEnables AI clients to generate over 26 chart types via MCP, automatically starting a private rendering service.271-
- AlicenseNot gradedqualityBmaintenanceEnables governed, agent-agnostic data exploration by allowing users to ask natural language questions through MCP-compatible agents, executing safe, permission-scoped queries against data sources and returning interactive charts.15 npmApache 2.0
- FlicenseNot gradedqualityBmaintenanceEnables agents to create interactive dashboards from natural language queries, returning both text results and portable dashboard resources for MCP Apps-compatible hosts.-
- AlicenseAqualityBmaintenanceEnables AI assistants to query Oracle AI Database read-only, profile tables, and render result sets as PNG charts.7MIT