wikipedia-mcp MCP — ユースケース, インストール & ライブデモ

なぜ使うのか

主な機能

10+ tools: search, article, summary, sections, links, coordinates, related topics
Multi-language with regional variants (zh-Hant, zh-Hans, sr-Latn, etc.)
Optional caching for repeat queries
Country/locale codes accepted alongside language codes
stdio, HTTP, streamable-HTTP transports

ライブデモ

実際の動作

wikipedia.replay ▶ 準備完了

0/0

インストール

クライアントを選択

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

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

Claude Desktop → Settings → Developer → Edit Config を開く。保存後、アプリを再起動。

~/.cursor/mcp.json · .cursor/mcp.json

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

Cursor は Claude Desktop と同じ mcpServers スキーマを使用。プロジェクト設定はグローバルより優先。

VS Code → Cline → MCP Servers → Edit

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

Cline サイドバーの MCP Servers アイコンをクリックし、"Edit Configuration" を選択。

~/.codeium/windsurf/mcp_config.json

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

Claude Desktop と同じ形式。Windsurf を再起動して反映。

~/.continue/config.json

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

Continue はマップではなくサーバーオブジェクトの配列を使用。

~/.config/zed/settings.json

{
  "context_servers": {
    "wikipedia": {
      "command": {
        "path": "uvx",
        "args": [
          "wikipedia-mcp"
        ]
      }
    }
  }
}

context_servers に追加。保存時に Zed がホットリロード。

claude mcp add wikipedia -- uvx wikipedia-mcp

ワンライナー。claude mcp list で確認、claude mcp remove で削除。

ユースケース

実用的な使い方： wikipedia-mcp

Cite Wikipedia instead of relying on model memory

👤 Anyone doing research-style Q&A ⏱ ~2 min beginner

使うタイミング： The model's training cutoff is older than the facts you need (new events, refreshed biographies, updated stats).

フロー

Ask with citation requirement

When did the 2026 Winter Olympics happen and where? Use the wikipedia MCP and cite the article.✓ コピーしました

→ Answer + Wikipedia article title/URL

結果： Answers with a verifiable source, less hallucination.

注意点

Wikipedia itself has errors / vandalism — For high-stakes facts, cross-check with a second source

Build a one-page dossier on a person or topic

👤 Journalists, prep-for-meeting folks ⏱ ~10 min beginner

使うタイミング： You're meeting someone unfamiliar or writing about a new-to-you topic and want structured context fast.

フロー

Summary first

Get a summary of [person/topic] from Wikipedia. Then list sections so I can pick which to deep-read.✓ コピーしました

→ Summary + TOC
Pull the key sections

Give me full text of the 'Career' and 'Controversies' sections.✓ コピーしました

→ Section text
Related threads

What related Wikipedia topics should I also skim?✓ コピーしました

→ Related topic list with 1-line summaries

結果： 10-minute dossier better than ad-hoc googling.

注意点

Wikipedia page doesn't exist for the person — Fall back to web search (firecrawl) for less-covered subjects

組み合わせ： notion

Compare how a topic is covered across languages

👤 Researchers, translators, international ops ⏱ ~15 min intermediate

使うタイミング： You suspect the English and Chinese Wikipedia articles on a topic differ materially (they often do for geopolitical topics).

フロー

Pull both versions

Get Wikipedia articles on [topic] in English and Simplified Chinese (zh-Hans).✓ コピーしました

→ Two article bodies
Diff the framing

What facts are in one but not the other? How does each frame the controversy?✓ コピーしました

→ Structured comparison

結果： Nuanced multi-perspective understanding.

注意点

Translation quality varies — Ask Claude to quote original language too for contested phrasings

組み合わせ

他のMCPと組み合わせて10倍の力を

wikipedia + notion

Build a 'places visited' Notion DB with Wikipedia summaries

For each row in my 'Trips' Notion DB, fetch the Wikipedia summary of the destination city and populate the 'About' column.✓ コピーしました

wikipedia + google-map

Show a topic on a map

Get coordinates of these 10 battles from Wikipedia; render them on a static Google map.✓ コピーしました

ツール

このMCPが提供する機能

ツール	入力	呼び出すタイミング	コスト
search_wikipedia	query: str, language?: str, limit?: int	Find candidate articles	free
get_article	title: str, language?: str	Need full content; use summary if 2-3 sentences suffice	free
get_summary	title, language?	Quick context without full article cost	free
get_sections	title, language?	Decide which sections to pull before pulling full article	free
get_section	title, section, language?	Cheaper than full article when you know what you want	free
get_links	title, language?	Discover related articles	free
get_coordinates	title	Plot an article's place on a map	free
get_related_topics	title	Research-mode expansion	free

コストと制限

運用コスト

APIクォータ: Wikipedia rate limits: ~200 requests/sec from an IP; add delays for long scripts
呼び出しあたりのトークン: Full articles 2k-15k tokens; summaries 200-500
金額: Free
ヒント: Always start with get_summary or get_sections — avoid pulling a whole get_article unless you need the full text.

セキュリティ

権限、シークレット、影響範囲

認証情報の保管： Optional Wikimedia OAuth token for higher rate limits

データ送信先： Outbound to Wikipedia's servers only

Wikipedia is crowd-edited — always treat as a starting point, not final source for contested facts.
Anonymous requests are throttled more aggressively; authenticate if you hit limits.

トラブルシューティング

よくあるエラーと対処法

403 Forbidden on heavy use

Anonymous rate limit hit. Provide a personal access token or slow down (add sleep).

Page not found

Titles are case-sensitive and disambiguation-specific. Use search_wikipedia first to get the canonical title.

Chinese article returns Traditional when I want Simplified

Pass language=zh-Hans or language=zh-Hant explicitly.

代替案

wikipedia-mcp 他との比較

代替案	代わりに使う場面	トレードオフ
DBpedia / Wikidata MCP	You want structured data (infobox fields) rather than prose	Better for querying facts; worse for reading
Direct Wikipedia API via fetch MCP	You need maximum control	You parse the JSON yourself

その他

リソース

📖 GitHub の公式 README を読む

🐙 オープンな issue を見る

🔍 400以上のMCPサーバーとSkillsを見る