titlize
Documentation

API Reference

Assistant Connector (MCP)

Use Titlize from AI assistants like Claude Code and Cursor through the Model Context Protocol connector.

2 min read

The Titlize assistant connector is a Model Context Protocol (opens in new tab) (MCP) server. It lets AI assistants — Claude Code, Cursor, and any other MCP-capable client — generate images and check your usage in natural language, without writing API calls by hand.

What the connector exposes#

The MCP server exposes two tools:

ToolWhat it does
generate_imageGenerates a titled image from a source image and a title. Supports the same template, aspect, and fileFormat options as the REST API.
check_usageReports your current plan, monthly quota, and how many images you have left.

The connector talks to the same API as everything else, so it respects your plan limits and authenticates with your API token.

Prerequisites#

  • Node.js 22 or newer
  • A Titlize API token (create one in the dashboard (opens in new tab))
  • An MCP-capable assistant (Claude Code, Cursor, etc.)

Install and configure#

The connector runs with npx — no install step required.

Claude Code#

Bash
1claude mcp add --transport stdio titlize \
2  --env IMAGE_TITLER_API_TOKEN=your_api_token \
3  --env IMAGE_TITLER_BASE_URL=https://api.titlize.com \
4  -- npx -y @image-titler/mcp-connector

Manual configuration#

Add the server to your assistant's MCP config:

JSON
1{
2  "mcpServers": {
3    "titlize": {
4      "command": "npx",
5      "args": ["-y", "@image-titler/mcp-connector"],
6      "env": {
7        "IMAGE_TITLER_API_TOKEN": "your_api_token",
8        "IMAGE_TITLER_BASE_URL": "https://api.titlize.com"
9      }
10    }
11  }
12}

Using it#

Once configured, ask your assistant in plain language, for example:

Generate a square share image titled "Ten ways to make your writing land" from hero.jpg.

The assistant calls generate_image with the right parameters and returns the finished image URL. Ask it to "check my Titlize usage" to call check_usage.

Next Steps#