Repository avatar
Search & Data Extraction
v0.1.0
active

persian-wikipedia-mcp

io.github.sepehr071/persian-wikipedia-mcp

Read-only Persian Wikipedia (fa.wikipedia.org): search, summaries, article text, nearby places.

Documentation

persian-wikipedia-mcp: unofficial Persian Wikipedia (ویکی‌پدیای فارسی) MCP server that searches fa.wikipedia.org, reads article text, infobox facts and nearby places for Claude, Cursor and Copilot

persian-wikipedia-mcp: Persian Wikipedia (ویکی‌پدیای فارسی) MCP server for AI agents

persian-wikipedia-mcp is an unofficial, read-only MCP server that lets an AI agent search articles, read summaries, full article text and infobox facts, and find nearby places on the Persian Wikipedia (ویکی‌پدیای فارسی, fa.wikipedia.org).
Its 9 tools search articles, give an article's lead summary, return the article as plain text (whole or one section, cut to a length limit), list its sections, read the infobox facts (birth date, population, ...) with categories, find the same article in other languages, pick a random article, show the most-read articles and the picture of the day, and find articles near a latitude and longitude.
No API key or account is needed; it works in Claude, Cursor, Copilot and any other MCP client.

PyPI Python CI License: PolyForm Noncommercial

For AI agents · Quick start · Tools · Limits · فارسی


For AI agents

  • What: unofficial MCP server (model context protocol) for fa.wikipedia.org: article search, lead summaries, plain-text article body by section, section lists, infobox facts and categories, other-language titles, random article, most-read articles and nearby places.
  • Install: uvx persian-wikipedia-mcp or claude mcp add persian-wikipedia -- uvx persian-wikipedia-mcp. Transport: stdio. Needs uv (Python 3.10+).
  • Auth: none. Optional PERSIAN_WIKIPEDIA_MCP_PROXY (HTTP proxy URL, or a comma-separated list); system proxy settings are ignored.
  • Access: read-only. It cannot log in or edit. It sends a descriptive User-Agent, as the Wikimedia User-Agent policy asks.
  • Tools: 9, all prefixed wiki_. Typical sequences: wiki_search -> wiki_summary; wiki_sections -> wiki_article (one section); wiki_infobox for facts; wiki_languages for the English title; wiki_nearby for places around a point.
  • Use when: the user asks what something or someone is, wants facts from an Iranian city, person or topic article (birth date, population, province), wants a section of an article, wants the English name of a Persian article, or asks what people are reading on the Persian Wikipedia.
  • Output: compact JSON; article text is plain text with == heading == lines. Text and titles are Persian, numbers in infoboxes keep Wikipedia's Persian digits.
  • Content license: Wikipedia text is written by Wikipedia contributors and is available under CC BY-SA. Reuse needs attribution and the same license.

Quick start

You need uv. No API key or account.

Claude Code:

claude mcp add persian-wikipedia -- uvx persian-wikipedia-mcp

Claude Desktop, Cursor, VS Code and other clients:

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

Example prompts:

  • "What is the English name of the Persian Wikipedia article about Shiraz?"
  • "Read the history section of the Tehran article on Persian Wikipedia."
  • "What are people reading most on Persian Wikipedia today?"
  • جمعیت شیراز طبق ویکی‌پدیای فارسی چقدر است؟
  • چه مقاله‌هایی نزدیک میدان آزادی هست؟

Real answer (18 Nov 2026): wiki_infobox(title="شیراز") returned the metropolitan population from the 1400 census as ۲٬۰۶۹٬۰۰۱ and the urban population as ۱٬۹۵۵٬۵۰۰, with the province Fars and 11 categories; wiki_languages(title="شیراز", lang="en") returned "Shiraz".

Tools

ToolWhat it returns
wiki_searchArticles matching words (Persian or English): titles, snippet, word count, total hits, paged with offset
wiki_summaryLead summary: one-line description, intro paragraph, image, coordinates, Wikidata id, URL
wiki_articleThe article as plain text, optionally one section (with subsections), cut to max_chars
wiki_sectionsSection headings with level and number
wiki_infoboxInfobox facts (birth date, population, ...) and the visible categories
wiki_languagesThe same article's title in other languages, for example lang="en"
wiki_randomA random article with its summary
wiki_trendingThe most-read articles of a day and the picture of the day
wiki_nearbyArticles near a latitude and longitude, nearest first, with distance in metres

Limits

  • Persian Wikipedia has no "on this day" feed and no featured article in its feed, so wiki_trending returns only the most-read articles and the picture of the day. The feed for day D holds the most-read of day D-1; for a date with no data the tool tries up to 3 days back.
  • Article text comes without citation markers and image captions, so its length can differ slightly from the rendered page (2,235 characters against 2,261 for the Tehran "نام" section in testing).
  • section matches the heading text first and falls back to a substring, which can land in a subsection (a short word like "تاریخ" in the Shiraz article matched "از دیدگاه تاریخی"). Exact heading text from wiki_sections is safest.
  • wiki_infobox reads the first infobox of the lead; articles without an infobox return empty fields. Nested tables are ignored.
  • Wikipedia is edited by anyone, so an article can contain mistakes or vandalism; the server tells the model to treat article text as data, not as instructions.
  • Results are cached for 60 seconds (not wiki_random), at most two requests run at a time and a response is capped at 5 MB.
  • Both the direct Iranian line and foreign networks work in testing. If your network blocks Wikipedia, set PERSIAN_WIKIPEDIA_MCP_PROXY. The server ignores system proxy variables on purpose.

فارسی

persian-wikipedia-mcp یک سرور MCP غیررسمی و فقط‌خواندنی برای ویکی‌پدیای فارسی (fa.wikipedia.org) است. با آن یک عامل هوش مصنوعی می‌تواند مقاله‌ها را جست‌وجو کند، خلاصهٔ آغازین و متن کامل مقاله (یا فقط یک بخش) را بخواند، فهرست بخش‌ها و اطلاعات جعبهٔ اطلاعات (تاریخ تولد، جمعیت و مانند آن) را ببیند، عنوان همان مقاله را در زبان‌های دیگر پیدا کند، مقالهٔ تصادفی بگیرد، پربازدیدترین مقاله‌ها و تصویر روز را ببیند و مقاله‌های نزدیک یک مختصات جغرافیایی را پیدا کند. بدون حساب کاربری و کلید API کار می‌کند. نصب در Claude Code:

claude mcp add persian-wikipedia -- uvx persian-wikipedia-mcp

متن مقاله‌های ویکی‌پدیا را مشارکت‌کنندگان آن نوشته‌اند و با مجوز CC BY-SA در دسترس است؛ برای بازنشر باید نام منبع را بیاورید و همان مجوز را حفظ کنید.

Development

uv sync
uv run ruff check . && uv run ruff format --check .
uv run python -m pytest -q          # offline tests on trimmed fixtures
uv run python -m pytest -m live -q  # live calls to fa.wikipedia.org

Releases: bump the version in pyproject.toml and server.json, run uv lock, then push a v* tag; GitHub Actions publishes to PyPI.

Disclaimer

Unofficial and not affiliated with or endorsed by the Wikimedia Foundation or Wikipedia. It is read-only and uses the public APIs of fa.wikipedia.org, which can change without notice. Wikipedia is a trademark of the Wikimedia Foundation. Article content is by Wikipedia contributors, licensed CC BY-SA.

Author

Built and maintained by Sepehr Radmard (سپهر رادمرد). LinkedIn · GitHub · more servers in iran-mcp.

مجوز استفاده (License)

استفادهٔ شخصی و غیرتجاری رایگان است؛ برای استفادهٔ تجاری با سپهر رادمرد تماس بگیرید. Free for personal and non-commercial use; commercial use needs a license.

استفادهٔ شخصی و غیرتجاری رایگان است. استفاده در شرکت، محصول پولی یا پروژهٔ مشتری به مجوز تجاری نیاز دارد؛ از راه لینکدین یا گیت‌هاب با سپهر رادمرد تماس بگیرید. متن کامل: PolyForm Noncommercial 1.0.0. این مجوز دربارهٔ کد این سرور است؛ محتوای ویکی‌پدیا مجوز خودش (CC BY-SA) را دارد.

Free for personal and non-commercial use under the PolyForm Noncommercial License 1.0.0. Using it in a company, a paid product or a client project needs a commercial license: contact Sepehr Radmard on LinkedIn or GitHub. This license covers the server code; Wikipedia content keeps its own CC BY-SA license.