MediaWiki MCP server
A maintained MCP server that connects an agent to any MediaWiki — including Wikipedia — to search, read, create, and edit pages.
MediaWiki MCP Server is an actively maintained server from Professional Wiki that lets an AI client interact with any MediaWiki wiki, from a private company wiki to Wikipedia itself. On the read side an agent can fetch a page or up to fifty pages at once, search by full text or title prefix, list a page's revision history and a specific revision, diff two versions (by revision, title, or supplied wikitext), parse wikitext to HTML without saving, enumerate category members and the pages linking to a page, read recent changes across the wiki, fetch file pages and their image bytes, and pull site info or the current session identity. That makes it a strong grounding and research tool over curated wiki knowledge.
With OAuth authentication it adds full write capability: create, update, move, delete, and undelete pages, and upload files from disk or a URL — so you can ask the agent to draft or revise an article and have it save the change on your behalf. It supports configuring multiple wikis (add, remove, and list them) and includes helpers for OAuth status and logout, plus optional tools for the Semantic MediaWiki, Cargo, and Bucket extensions when a wiki has them installed. The server is published as @professional-wiki/mediawiki-mcp-server, runs over stdio via npx, and can also be deployed over HTTP with a published container image. It is MIT-licensed.
Quick install
Copy-paste configs are provided for all 8 supported clients. Pick your client below.
Available tools
| Tool | Description |
|---|---|
| get-page | Fetches a single wiki page. |
| get-pages | Fetches multiple wiki pages in one call (up to 50). |
| get-revision | Fetches a specific page revision. |
| get-page-history | Lists a page's recent revisions. |
| compare-pages | Diffs two versions of a page by revision, title, or supplied wikitext. |
| parse-wikitext | Renders wikitext to HTML without saving. |
| search-page | Searches page titles and contents. |
| search-page-by-prefix | Searches page titles by prefix. |
| get-category-members | Lists the members of a category (up to 500, paginated). |
| get-links-here | Lists pages that link to a given page (up to 500, paginated). |
| get-recent-changes | Lists recent changes across the wiki (up to 50, paginated). |
| get-file | Fetches a file page. |
| get-file-data | Returns a file's image bytes as base64 for analysis. |
| get-site-info | Returns wiki settings, version, extensions, and license. |
| whoami | Reports the current session identity and group memberships. |
| create-page | Creates a new page (requires authentication). |
| update-page | Updates an existing page (requires authentication). |
| move-page | Renames (moves) a page (requires authentication). |
| delete-page | Deletes a page (requires authentication). |
| undelete-page | Restores a deleted page (requires authentication). |
| upload-file | Uploads a file from disk (requires authentication). |
| upload-file-from-url | Uploads a file from a URL (requires authentication). |
| update-file | Uploads a new revision of a file from disk (requires authentication). |
| update-file-from-url | Uploads a new revision of a file from a URL (requires authentication). |
| list-wikis | Lists all configured wikis (disabled when fewer than two are set up). |
| add-wiki | Adds a wiki by URL. |
| remove-wiki | Removes a configured wiki. |
| oauth-status | Lists stored OAuth tokens with their scopes and expiry. |
| oauth-logout | Removes stored OAuth tokens. |
| smw-query | Runs a Semantic MediaWiki #ask query (when SMW is installed). |
| smw-list-properties | Lists Semantic MediaWiki properties (when SMW is installed). |
| cargo-query | Runs a Cargo SQL-style query (when the Cargo extension is installed). |
| cargo-list-tables | Lists Cargo tables (when the Cargo extension is installed). |
| cargo-describe-table | Lists a Cargo table's fields (when the Cargo extension is installed). |
| bucket-query | Runs a Bucket extension Lua query (when the Bucket extension is installed). |
Required configuration
- CONFIGOptional
Path to the JSON config file naming the target wiki(s) and optional OAuth credentials.
What you can do with it
Ground answers in a wiki
The agent searches and reads pages from Wikipedia or your internal wiki, diffs revisions, and cites the source so its response rests on curated, verifiable knowledge.
Draft and save edits
With OAuth configured, ask the agent to create or revise an article — for example, build a page from the contents of a Wikipedia article — and it saves the change on your behalf.
FAQ
- Is it free?
- Yes. The server is open source under the MIT license and free to run. Reading public wikis needs no account; you only need credentials for write actions or for reading private wikis.
- Does it support remote/OAuth?
- Yes. It authenticates via OAuth 2.0, which is required for all write actions and for reading private wikis, and supports per-request bearer tokens in HTTP mode. It can run over stdio locally or be deployed over HTTP using the published container image.
- Can it edit Wikipedia?
- Yes. Point it at any MediaWiki wiki, including Wikipedia, and with authenticated OAuth it can create, update, move, delete, and undelete pages and upload files, subject to that wiki's permissions and editing policies.