WordPress MCP server for your AI

Seven modules: posts, pages, media, comments, taxonomies, users and site settings, which comes to 35 operations, 17 read and 18 write. Sixteen load into your AI client by default and the rest stay searchable, so the catalog does not crowd out your other connections. The write side is not just publishing: it includes trashing posts and pages, deleting comments, media items and categories, and updating site settings.

No. The connect form asks for three values, Site URL, WordPress username and Application password, and the application password is a separate credential you generate in WordPress rather than the password you sign in with. Because the site address is part of the credential, one connection covers one site.

Whichever user the application password belongs to. Run get_current_user if you are ever unsure, since it returns exactly the account the application password authenticates as. If you want an agent that only reports, pair a dedicated account with the read-only access mode rather than relying on prompt wording.

No. The media tools cover listing items, fetching one by id with get_media, updating a media item's metadata with update_media and deleting one, but there is no upload operation, so files still get added the way your team adds them today. What the agent can do is work after the fact: Media uploaded is one of the events PopMCP watches, and update_media can retitle or re-caption an item once it is there.

Most of the catalog is id-based, so get_post, update_post, delete_page and their siblings all expect an id rather than a title. The agent gets there with search_content, which searches site-wide across posts, pages and other public content, or with list_posts and list_pages, then acts on the ids that come back. If you already know the id, say it in the prompt and the lookup step disappears.

Post published, Draft created, Page created, Comment posted, Comment awaiting moderation, Media uploaded, Category created, Tag created and User registered. Post published is polled and the rest work by change detection, so PopMCP spots the new record itself instead of waiting on a webhook from your site. Comment awaiting moderation is the narrow one: it fires when a comment is held for approval instead of going live, not on every new comment.

No. PopMCP hosts it and gives you a Streamable HTTP MCP URL. There is nothing to deploy, no callback URL to register, and no infrastructure to keep running or patch.

Any client that supports a remote Streamable HTTP MCP server — Claude Desktop, Claude Web, ChatGPT custom connectors, Codex and the Codex CLI, OpenCode and Z.ai's ZCode, as well as editors such as Cursor, Windsurf and Zed. Your client signs in to PopMCP with OAuth; the provider credential never leaves the server.

Loading 35 tool definitions into every conversation burns context before the work starts. PopMCP loads a curated 16-tool set covering the everyday workflows and keeps the rest searchable, so your AI pulls a specialist tool only when a task needs it. You can switch a connection to the full catalog at any time.

A tool is something your AI calls — there are 35 here. A trigger is something PopMCP watches for you: 9 WordPress events that can start work on the PopMCP side. Triggers never enter your AI client's context, and none of them ask you to paste a callback URL into WordPress.