WordPress MCP Server

What is WordPress MCP Server?

It is an MCP (Model Context Protocol) server that enables publishing content to WordPress sites. It supports publishing articles, managing categories and tags, adding featured images, and testing connections. The server can run as both a standalone FastAPI application and an MCP server.

How to use WordPress MCP Server?

Set up a virtual environment with UV and install dependencies from requirements.txt. Configure WordPress credentials (site URL, username, application password) in a .env file. Run python main.py --mcp for MCP mode or python main.py for the standalone API. The MCP tools available are PUBLISH_ARTICLE, PREPARE_ARTICLE_METADATA, and TEST_CONNECTION.

Key features of WordPress MCP Server

• Publish articles to WordPress sites
• Test WordPress connection
• Create, retrieve, and manage categories and tags
• Automatically handle category and tag creation when publishing
• Support featured images via URL or base64 data
• Works as both standalone FastAPI server and MCP server

Use cases of WordPress MCP Server

• Automating article publishing from an AI assistant to a WordPress blog
• Managing WordPress taxonomy (categories and tags) programmatically
• Batch uploading articles with featured images
• Testing connectivity and credentials for a WordPress site
• Integrating WordPress publishing into MCP‑based workflows

FAQ from WordPress MCP Server

What are the required environment variables?

WP_SITE_URL, WP_USERNAME, and WP_PASSWORD must be set in a .env file. These serve as defaults for all tools.

How do I publish an article with a featured image?

Provide an image_url (direct link) or image_base64 (base64‑encoded data) in the PUBLISH_ARTICLE tool call. An optional image_filename sets a custom filename.

What MCP tools are available?

Three tools: PUBLISH_ARTICLE, PREPARE_ARTICLE_METADATA, and TEST_CONNECTION.

Does the server create categories and tags automatically?

Yes. The PREPARE_ARTICLE_METADATA tool checks for existing categories/tags and creates any that are missing, returning their IDs for use with PUBLISH_ARTICLE.

Can I run this as a standalone API?

Yes. Run python main.py without the --mcp flag to start a FastAPI server at http://localhost:8000.