Integrations

Plotivy MCP server: scientific figures inside your AI assistant

Plotivy exposes its figure engine over the Model Context Protocol (MCP). Connect it once and any MCP client — Claude, ChatGPT, Cursor, and others — can turn the data you are discussing into a publication-ready figure, returning both the rendered image and the reproducible Python code, with journal styles applied on request.

What the server provides

• create_figure

  Send CSV data and a plain-language description; get back the rendered figure (PNG) plus the exact Python matplotlib or Plotly code that produced it, with an optional journal style applied.

• render_figure_code

  Execute matplotlib or Plotly code you already have in Plotivy's sandbox and get the rendered figure, optionally against CSV data exposed as a DataFrame.

• list_journal_styles

  List the journal figure styles Plotivy can apply (Nature, Science, IEEE, and more). Free to call.

Step 1 — Create an API token

In your profile settings, open the Developer API Tokens section and generate a token. It starts with plt_ and is shown once — store it like a password. The same token also works for the Plotivy Chrome and Office integrations, and can be revoked at any time.

Step 2 — Connect your client

Claude Code

claude mcp add --transport http plotivy \
  https://plotivy-backend-production.up.railway.app/mcp \
  --header "Authorization: Bearer plt_YOUR_TOKEN"

Claude Desktop, Cursor, and other JSON-config clients

Add the server to your client's MCP configuration (for example .mcp.json in Cursor, or Settings → Connectors in Claude):

{
  "mcpServers": {
    "plotivy": {
      "type": "http",
      "url": "https://plotivy-backend-production.up.railway.app/mcp",
      "headers": {
        "Authorization": "Bearer plt_YOUR_TOKEN"
      }
    }
  }
}

ChatGPT

In ChatGPT, custom MCP connectors are available in developer mode (Settings → Connectors → Advanced). Add a connector with the server URL above and your plt_ token as the bearer authentication header.

Step 3 — Ask for a figure

Paste or reference tabular data in your conversation and ask, for example: "Use Plotivy to make a violin plot of expression by treatment group in Nature style." The assistant calls create_figure and returns the figure, the Python code, and a link to keep editing the figure in the Plotivy workspace.

Limits and privacy

• Figure generation shares your account's daily chart quota (the same one the web app uses). list_journal_styles is free.
• Data sent to the server is processed in memory to render the figure and is not used to train models.
• CSV payloads are capped at 512 KB / 5,000 rows per call — aggregate or sample larger datasets first.
• Tokens can be revoked instantly in profile settings; revocation takes effect within a minute.