MCP Directory

Wikipedia MCP

Give your LLM live Wikipedia access: search, full articles, summaries, sections, links and key facts in any language.

Unverified
stdio (local)
No auth
Python
View repo 270

Add to your client

Copy the config for your MCP client and paste it into its config file.

Install / run
pipx install wikipedia-mcp

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "wikipedia-mcp": {
      "command": "wikipedia-mcp",
      "args": []
    }
  }
}

Step-by-step guides: Add to Claude Desktop · Add to Cursor · Add to Windsurf

Before you start

  • Python 3 with pipx (recommended) or pip
  • No API key required; optional free Wikipedia access token to avoid rate limits
  • An MCP client such as Claude Desktop

About Wikipedia MCP

The server wraps Wikipedia's public API and returns structured JSON (titles, snippets, sections, links, categories) instead of raw HTML, so tool results stay parseable. Every tool also ships a wikipedia_-prefixed alias (e.g. wikipedia_get_article) to stay discoverable when many MCP servers are connected at once.

Its strongest feature for agent use is token economy: instead of dumping a full article into context, summarize_article_for_query returns a summary focused on your question, summarize_article_section targets one section, and extract_key_facts returns N bullet facts. get_coordinates adds latitude/longitude for places, which is handy in mapping or travel workflows.

Language handling is unusually thorough. --language takes raw codes including variants (zh-hans, zh-tw, sr-latn, ku-arab), while --country maps 140+ intuitive codes like US, Taiwan or Germany to the right Wikipedia edition; you can even register several servers (wikipedia-us, wikipedia-japan) side by side in one config.

Trade-offs: very large articles can exceed token limits (prefer summaries and sections), and Wikipedia rate-limits heavy traffic — a free personal access token lifts that. There is a built-in test_wikipedia_connectivity diagnostic tool, and search responses include status metadata so empty results are debuggable rather than silent.

Tools & capabilities (11)

search_wikipedia

Search Wikipedia for articles matching a query, with result metadata and status.

get_article

Get full article content: text, summary, sections, links and categories.

get_summary

Get a concise summary of an article.

get_sections

Get a structured list of an article's sections with content.

get_links

Get the links contained within an article.

get_coordinates

Get latitude/longitude coordinate data for an article.

get_related_topics

Get related topics based on an article's links and categories.

summarize_article_for_query

Summarize an article focused on a specific query.

summarize_article_section

Summarize one section of an article.

extract_key_facts

Extract N key facts from an article, optionally scoped to a topic.

test_wikipedia_connectivity

Diagnostic: check reachability of Wikipedia's API with response-time info.

What this server can do

Wikipedia MCP provides tools for these capabilities — tap one to see every MCP server that does the same:

Search the webScrape web pages

When to use it

  • Ground answers with sources: search a topic, then pull the exact section instead of trusting model memory
  • Fact-check a claim by extracting key facts from the relevant article
  • Query non-English Wikipedia (--country Japan, --language zh-tw) for region-specific coverage
  • Fetch coordinates of places for mapping or travel-planning workflows

Quick setup

  1. 1pipx install wikipedia-mcp (pipx keeps the command on Claude Desktop's PATH)
  2. 2Add a wikipedia entry with command wikipedia-mcp under mcpServers in claude_desktop_config.json
  3. 3If the client reports spawn ENOENT, run which wikipedia-mcp and use the full path as the command
  4. 4Optional: add args like --country TW, --enable-cache or --access-token <token>
  5. 5Restart Claude Desktop and ask: What does Wikipedia say about quantum computing?

Security notes

Read-only against Wikipedia's public API — no account, no key, no private data, and it writes nothing to your system beyond optional caching. If you expose the http/streamable-http transport beyond localhost, enable --auth-mode static or jwt so the endpoint is not open to everyone.

Wikipedia MCP FAQ

Is the Wikipedia MCP free to use?

Yes. It is MIT-licensed, needs no API key, and talks to Wikipedia's free public API. Heavy usage can hit rate limits (403 errors); a free Wikipedia personal access token passed via --access-token removes that throttle.

Claude Desktop says spawn wikipedia-mcp ENOENT — what is wrong?

The command is not on Claude Desktop's PATH. Either install with pipx (which puts it in a global location) or set the config command to the absolute path printed by which wikipedia-mcp.

Can I use it with non-English Wikipedia?

Yes — pass --language (ja, zh-hans, zh-tw, sr-latn, ...) or the friendlier --country flag (Taiwan, Japan, Germany, ...). It maps 140+ countries and regions to the right language edition, including Chinese simplified/traditional variants, and you can run several language-specific server entries at once.

#wikipedia#knowledge#search#research#grounding

Alternatives to Wikipedia MCP

Compare all alternatives →
Fetch (Reference)
Search & Web
87k

Official MCP reference server that fetches a URL and returns its content as clean Markdown, with chunking.

Verified
stdio (local)
No auth
Python
2 tools
Updated 25 days agoRepo
Firecrawl
Search & Web
6.6k

Official Firecrawl MCP server — scrape, crawl, map, search, and structured extraction for any LLM client.

Verified
stdio (local)
API key
TypeScript
8 tools
Updated 23 days agoRepo
Exa Search
Search & Web
4.6k

Exa's neural web search and crawling MCP server — runs locally via npx or as a hosted remote endpoint.

Verified
stdio (local)
API key
TypeScript
6 tools
Updated 24 days agoRepo

Compare Wikipedia MCP with:

vs Fetch (Reference)vs Firecrawlvs Exa Searchvs ArXiv MCP Server