ya-reviews-mcp

ya-reviews-mcp

MCP server that scrapes Yandex Maps business reviews and company info, exposing them to AI assistants via three tools. No API key required, uses headless browser.

Category
访问服务器

README

ya-reviews-mcp

PyPI License Python FastMCP

Model Context Protocol (MCP) server for Yandex Maps business reviews. Scrapes reviews from any Yandex Maps organization page and exposes them to your AI assistant — review text, ratings, author info, likes/dislikes, business responses, and company metadata.

No API key required. Uses a headless Chromium browser under the hood. Supports multiple browser backends: Playwright, Patchright (anti-detection), or a remote browser via CDP.

Quick Start

1. Install

Choose your backend:

# Option A: Playwright (default)
uv tool install ya-reviews-mcp[playwright]
playwright install chromium

# Option B: Patchright (better anti-detection)
uv tool install ya-reviews-mcp[patchright]
patchright install chromium

# Option C: Remote browser (no local browser needed)
uv tool install ya-reviews-mcp
# Then run a browser separately, e.g.:
# docker run -p 3000:3000 ghcr.io/browserless/chromium

No uv? Use pip instead: pip install ya-reviews-mcp[playwright]

2. Find the Organization ID

Open any business on Yandex Maps and look at the URL:

https://yandex.ru/maps/org/astra_motors/1248139252/reviews/
                                        ^^^^^^^^^^
                                        this is the org_id

3. Configure Your IDE

Add to your Claude Desktop / Cursor MCP configuration:

{
  "mcpServers": {
    "ya-reviews": {
      "command": "uvx",
      "args": ["--from", "ya-reviews-mcp[playwright]", "ya-reviews-mcp"]
    }
  }
}

Using Patchright? Replace ya-reviews-mcp[playwright] with ya-reviews-mcp[patchright] and add "--backend", "patchright" to args.

Running from source? Use uv run instead:

{
  "mcpServers": {
    "ya-reviews": {
      "command": "uv",
      "args": ["run", "--directory", "/path/to/ya-reviews-mcp", "ya-reviews-mcp"]
    }
  }
}

4. Start Using

Ask your AI assistant to:

  • "Get reviews for organization 1248139252" — fetch all reviews with text, ratings, and business responses
  • "Show me company info for org 1248139252" — company name, rating, review count
  • "Give me a summary of org 1248139252 with the last 10 reviews" — company info + recent reviews in one call

Tools

3 tools for Yandex Maps review data:

Tool Description
get_reviews Fetch reviews with text, author, date, rating, likes/dislikes, direct review URL, and business responses. Supports max_reviews limit and sort (by_time or by_rating).
get_company_info Company metadata: name, rating, review count, address, categories.
get_company_summary Company info + recent reviews in one call. Best for a quick business overview.

Review Data Fields

Each review includes:

Field Description
author_name Reviewer's display name
author_icon_url Avatar image URL
author_profile_url Link to the reviewer's Yandex Maps profile
date ISO 8601 date string
text Full review text
stars Rating (1.0 - 5.0)
likes Thumbs up count
dislikes Thumbs down count
review_url Direct link to this specific review
business_response Organization's reply text (if any)

Company Info Fields

Field Description
name Business name
rating Overall rating (e.g., 4.8)
review_count Total number of reviews
address Business address
categories Business categories list

Configuration

All configuration via environment variables (no API key needed):

Variable Default Description
BROWSER_BACKEND playwright Browser backend: playwright, patchright, or remote
BROWSER_WS_URL WebSocket URL for remote backend. Short form ws://host:port auto-discovers the full debugger URL.
BROWSER_HEADLESS true Set to false for visual debugging
PAGE_TIMEOUT 30000 Page load timeout in ms
INTERCEPT_TIMEOUT 15000 Max wait for reviews to appear in DOM (ms)
REQUEST_DELAY 2.0 Delay between scroll loads (seconds)
MAX_PAGES 20 Max scroll iterations (50 reviews per scroll)
SCRAPER_RETRIES 3 Retry attempts for page loads
SCRAPER_RETRY_DELAY 2.0 Base delay between retries (seconds)
BROWSER_LOCALE ru-RU Browser locale
ENABLED_TOOLS all Comma-separated list of allowed tools

Copy .env.example to .env and uncomment the values you want to change.

CLI

# stdio (default, for MCP clients)
ya-reviews-mcp

# HTTP transport
ya-reviews-mcp --transport streamable-http --port 8000

# With verbose logging
ya-reviews-mcp -vv

# Use Patchright backend
ya-reviews-mcp --backend patchright

# Use remote browser
ya-reviews-mcp --backend remote --browser-url ws://localhost:3000

# Load custom .env file
ya-reviews-mcp --env-file /path/to/.env

How It Works

Unlike ya-metrics-mcp which uses the official Yandex Metrika API, Yandex Maps has no public API for reviews. This server uses a headless browser to:

  1. Navigate to the organization's reviews page on yandex.ru/maps
  2. Parse review data from the server-rendered DOM (schema.org structured data)
  3. Scroll to load additional reviews (50 per scroll, infinite pagination)
  4. Click "Show organization response" buttons to reveal business replies
  5. Extract all fields including likes/dislikes and direct review URLs

A single Chromium browser instance is shared across requests via FastMCP's lifespan context. Each scrape creates a fresh browser context (isolated cookies/state).

Development

# Install with dev dependencies (includes playwright)
uv sync --extra dev

# Install browser
uv run playwright install chromium

# Run tests
uv run pytest

# Lint
uv run ruff check src/

# Type check
uv run mypy src/

License

MIT

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选