Skip to main content
Glama
zhugejun
by zhugejun
README.md
# @chartone/mcp

> Branded chart rendering for AI agents. Hosted MCP, zero setup.

ChartOne MCP gives any AI agent the ability to produce themed charts and return them as both an inline image and a persistent hosted URL — perfect for emails, Slack messages, PDFs, and dashboards.

## Install

### Claude Desktop

Add to `~/Library/Application Support/Claude/claude_desktop_config.json` (macOS) or `%APPDATA%\Claude\claude_desktop_config.json` (Windows):

```json
{
  "mcpServers": {
    "chartone": {
      "command": "npx",
      "args": ["-y", "@chartone/mcp"],
      "env": {
        "CHARTONE_API_KEY": "sk-..."
      }
    }
  }
}
```

Restart Claude Desktop. Anonymous mode (no key) returns watermarked charts.

### Cursor / Zed / Windsurf

Same config shape, in each client's MCP settings file. See [docs.chartone.dev/mcp](https://docs.chartone.dev/mcp) for current paths.

### Hosted endpoint (beta)

For cloud agents that can't run local stdio servers:

```text
POST https://api.chartone.dev/mcp
Authorization: Bearer sk-...   (optional, anonymous returns watermark)
Content-Type: application/json

Body: { "method": "tools/list" }  or  { "method": "tools/call", "params": {...} }
```

## Tools

- **`render_chart`** — render a chart from explicit type, data, theme.
- **`suggest_chart`** — give it data + intent ("show quarterly trends"), it picks the chart type/theme and renders.

Both return an inline image plus a persistent hosted URL.

## Get an API key

Sign up at [chartone.dev](https://chartone.dev). Free tier includes a generous render quota; paid tiers remove the watermark and extend URL lifetime.

TDQS

B3.4/5.0

Scored across 2 tools

Disambiguation4/5

Both tools generate charts, but render_chart takes explicit data and renders a chart, while suggest_chart uses AI to recommend chart type and theme based on intent. The descriptions clearly differentiate their roles, though there is minor overlap in that both produce chart images.

Naming Consistency5/5

Both tool names follow a consistent verb_noun pattern using snake_case (render_chart, suggest_chart), making them predictable and easy to understand.

Tool Count4/5

With only 2 tools, the server feels minimal but appropriately scoped for a focused chart rendering and suggestion utility. The count is slightly under what might be expected, but each tool has a clear purpose.

Completeness3/5

The server covers chart creation and intelligent suggestion, but lacks tools for managing or editing previously rendered charts (e.g., update, delete, list). This is a notable gap if charts are stored with persistent URLs.

Maintenance

ActivityInactive
ResponsivenessNo issues