Text-Image MCP Server

A lightweight Model Context Protocol (MCP) server built with Python to allow agentic AI applications to quickly generate stylized imagery from text and code.

Features

• Pygments Code Highlighting: Automatically formats programming languages (JSON, Python, HTML, bash, etc.) using intelligent token mapping and syntax-coloring right out of the box.

• Agent-Driven Markdown (English Processor): Native lightweight parser detects unstructured text and allows agents to explicitly define importance by parsing **bold** strings and `inline code blocks` independently.

• Embedded Typography: Includes bundled instances of Fira Sans and Fira Mono to produce stunning desktop-grade visual representation.

Related MCP server: openai-gpt-image-1-mcp

Usage

This project wraps into the default MCP communication streams gracefully. Agents interacting with the API via the generate_text_image tool can inject the following parameters:

{
  "text": "Your markdown or code here",
  "language": "markdown", 
  "bg_color": [20, 24, 30],
  "text_color": [220, 230, 240],
  "font_size": 24
}

Note: If language is omitted, the server will intelligently attempt to guess the incoming programming language.

Installation

# Clone the project and instantiate a virtual environment
python -m venv venv

# Windows
.\venv\Scripts\activate

# Linux/macOS
source venv/bin/activate

# Install the dependencies
pip install mcp Pillow Pygments

Running the Server

If you are using Claude Desktop, Cline, or Antigravity, you simply register the server utilizing the following configurations within your host's MCP environment:

{
  "mcpServers": {
    "text-image": {
      "command": "C:/path/to/project/venv/Scripts/python.exe", // Use /path/to/project/venv/bin/python on Linux/macOS
      "args": ["-m", "text_image_mcp"],
      "env": {
         "PYTHONPATH": "C:/path/to/project/src" // Use /path/to/project/src on Linux/macOS
      }
    }
  }
}

GitHub Copilot

For GitHub Copilot within VS Code, you can add the server by navigating to the command palette and typing MCP: Open User Configuration. This will open your user mcp.json file. Provide the identical configuration block shown above (ensuring you adjust paths for your specific OS). Alternatively, you can drop .vscode/mcp.json inside your project root directory.

Testing Protocol

100% of the image rendering boundary mapping logic is covered seamlessly through pytest suites mirroring actual Agentic constraints. Run tests using:

python -m pytest tests/

License

MIT License. See LICENSE for details.