KudaGo + Nominatim MCP Server

KudaGo + Nominatim MCP Server

MCP server combining KudaGo and Nominatim APIs with configurable stdio or streamable HTTP transport.

Category
访问服务器

README

KudaGo + Nominatim MCP Server

MCP server for discovering concerts, theater performances, exhibitions, lectures, festivals, movies, movie showings, venues and other leisure activities.

The server combines:

  • KudaGo Public API v1.4 for events, venues, movies, showings, news, lists and reference data.
  • OpenStreetMap Nominatim-compatible geocoding for resolving natural place names into coordinates.

It is designed as a convenient tool layer for AI agents: an agent can ask for events near a district, town, landmark or city name even when that place is not a native KudaGo location slug.

This project is an MVP. By default it uses the public Nominatim service with a 1 request per second local throttle (NOMINATIM_MIN_INTERVAL_SECONDS=1.0), which is roughly 60 geocoding requests per minute. That is useful for prototypes and personal workflows, but not enough for high-volume production traffic.

Features

  • Search events by KudaGo location slug, natural place name or explicit coordinates.
  • Search concerts, theater, exhibitions, lectures, festivals and free events via KudaGo categories and tags.
  • Search venues near a location or within a radius.
  • Search movies and movie showings.
  • Resolve cities, towns, districts and landmarks through Nominatim.
  • Detect ambiguous places instead of silently picking the first geocoding result.
  • Retrieve KudaGo news, curated lists, object details, comments and reference dictionaries.
  • Return structured JSON envelopes that are easy for AI agents to inspect.
  • Run as an MCP server over stdio or Streamable HTTP.

Architecture

AI agent / MCP client
        |
        v
FastMCP tools
        |
        v
Location-aware leisure search
        |
        +-- kudago_mcp_client / KudaGoHttpClient
        |   +-- events, places, movies, showings, news, lists, references
        |
        +-- kudago_nominatim_geo.resolve_geo_for_kudago
            +-- nominatim_geo_client / NominatimHttpClient
                +-- place resolution and ambiguity detection

The MCP layer exposes a small set of tools and normalizes common user intents. KudaGo is responsible for leisure data. Nominatim is responsible for resolving plain-language place names into coordinates when KudaGo does not have a direct location match.

Geo resolution follows this order:

  1. Use location directly when a KudaGo slug such as msk or spb is provided.
  2. Try to match place_query against KudaGo locations by slug or localized name. The KudaGo locations dictionary is cached in memory for 15 minutes by default.
  3. Use generic Nominatim search for cities, districts, addresses and landmarks, then pass the resulting lat, lon and radius to KudaGo endpoints that support coordinates.
  4. Return an explicit geo.status such as ambiguous_place, place_not_found or unsupported when the server cannot safely continue.

Example Queries

  • Find concerts in Moscow next week.
  • What theater performances are available tomorrow in Saint Petersburg?
  • What events are happening in Nakhabino?
  • I want to watch an action movie near TRK Shchuka.
  • Suggest free science and technology events this weekend.
  • Find museums or exhibition venues near this address.
  • Show movie showings in a specific cinema.

Available Tools

All tools return a JSON object with at least:

  • status: ok or error.
  • tool: the tool name.
  • data: the upstream KudaGo or Nominatim response when the call succeeds.
  • geo: geo resolution details for location-aware tools.

resolve_place

Resolve a natural place name through Nominatim.

Useful when an agent needs coordinates, wants to check ambiguity, or needs to ask the user which place they meant.

Parameters:

  • query: place name, address, district or landmark.
  • countrycodes: optional comma-separated country filter, defaults to ru.
  • limit: number of candidates from 1 to 10.
  • accept_language: response language, for example ru or en.

reference

Fetch KudaGo dictionaries.

Supported kind values:

  • event_categories
  • place_categories
  • locations
  • location
  • agent_roles
  • agent_role

Use this before category, place-category, location or role filtering when the agent needs exact KudaGo slugs or IDs.

search

Full-text KudaGo search.

Parameters include:

  • query: text query.
  • ctype: event, place, news or list.
  • location: KudaGo location slug.
  • place_query: natural place name resolved by KudaGo or Nominatim.
  • lat, lon, radius: explicit geo search.
  • is_free, include_inactual, expand, page, page_size, lang.

Use this for text search such as "jazz", "museum" or "where to go". For strict date/category filtering, prefer events.

events

Search KudaGo events with deterministic filters.

Parameters include:

  • location or place_query.
  • lat, lon, radius.
  • actual_since, actual_until: Unix timestamp or ISO 8601 date/time.
  • categories, tags: comma-separated KudaGo slugs.
  • is_free.
  • fields, expand, order_by, page, page_size, lang.

events_of_the_day

Fetch KudaGo editorial events of the day.

This tool requires a KudaGo location slug or a place_query that can be matched to a KudaGo location. It does not use Nominatim coordinates.

places

Search KudaGo venues and places.

Parameters include:

  • location or place_query.
  • lat, lon, radius.
  • categories, tags.
  • has_showings: useful for cinemas.
  • fields, expand, page, page_size, lang.

movies

Search KudaGo movies.

Parameters include:

  • location or place_query.
  • actual_since, actual_until.
  • premiering_in_location.
  • place_id, page, page_size, lang.

This tool works with KudaGo location slugs. It does not use Nominatim coordinates.

movie_showings

Search movie showings, show showings for a movie, or fetch one showing.

Modes:

  • Pass showing_id for /movie-showings/{id}/.
  • Pass movie_id for /movies/{id}/showings/.
  • Pass neither for /movie-showings/.

Parameters include:

  • location or place_query.
  • actual_since, actual_until.
  • place_id, is_free, page, page_size, lang.

This tool works with KudaGo location slugs. It does not use Nominatim coordinates.

news, lists, agents, object, comments

Additional KudaGo tools:

  • news: city news and fresh KudaGo materials.
  • lists: curated KudaGo lists and selections.
  • agents: people and organizations associated with KudaGo entities.
  • object: details for event, news, list, place, movie, movie_showing, agent, agent_role or location.
  • comments: comments for event, news, list, place or movie.

Ambiguous Locations

The server does not silently choose a location when multiple candidates exist.

Example input:

Alekseevka

Example response shape:

{
  "status": "error",
  "tool": "events",
  "message": "Several Nominatim candidates were found. Pick one and call the tool again with explicit lat, lon and radius.",
  "geo": {
    "status": "ambiguous_place",
    "kind": "none",
    "candidates": [
      {"display_name": "Alekseevka, Belgorod Oblast, Russia"},
      {"display_name": "Alekseevka, Samara Oblast, Russia"},
      {"display_name": "Alekseevka, Republic of Bashkortostan, Russia"}
    ]
  }
}

An agent should show the candidates to the user, ask for clarification, and then call the target tool again with explicit lat, lon and radius.

Installation

Clone the repository and install it into a Python 3.11+ environment:

git clone <repo-url>
cd kudago-nominatim-mcp-server
python -m venv .venv
.venv\Scripts\activate
python -m pip install -r requirements.txt

The dependencies in requirements.txt are sufficient to run the MCP server. For development and tests, install the project with its optional development dependencies:

python -m pip install -e ".[dev]"
python -m pytest

Running

Stdio is the default transport and is the best fit for most MCP desktop/client integrations:

python kudago_nominatim_mcp_server.py

The same command with an explicit transport:

python kudago_nominatim_mcp_server.py --transport stdio

Streamable HTTP:

python kudago_nominatim_mcp_server.py --transport http --host 127.0.0.1 --port 8010 --path /mcp/

Transport priority is:

CLI arguments > environment variables > defaults

MCP Configuration

Stdio example:

{
  "name": "kudago_nominatim",
  "alias": "kudago",
  "connect_type": "executable",
  "executable": "python",
  "args": ["/absolute/path/to/kudago_nominatim_mcp_server.py"],
  "env": {
    "KUDAGO_LANG": "ru",
    "NOMINATIM_USER_AGENT": "your-app-name/0.1.0 (your-email-or-url)",
    "NOMINATIM_EMAIL": "your-email@example.com",
    "NOMINATIM_MIN_INTERVAL_SECONDS": "1.0",
    "NOMINATIM_COUNTRYCODES": "ru",
    "KUDAGO_LOCATIONS_CACHE_TTL": "900"
  },
  "enabled": true
}

Streamable HTTP example:

{
  "name": "kudago_nominatim",
  "alias": "kudago",
  "connect_type": "streamable_http",
  "url": "http://127.0.0.1:8010/mcp/",
  "headers": {},
  "enabled": true
}

When using Streamable HTTP, start the server separately before starting the MCP client.

Configuration

Environment variables:

Variable Default Description
KUDAGO_BASE_URL https://kudago.com/public-api/v1.4/ KudaGo API base URL.
KUDAGO_LANG ru Default KudaGo response language.
NOMINATIM_BASE_URL https://nominatim.openstreetmap.org/ Nominatim-compatible API base URL.
NOMINATIM_USER_AGENT kudago-nominatim-mcp/0.1.0 Application User-Agent sent to Nominatim. Set a real contact value.
NOMINATIM_REFERER empty Optional Referer header for Nominatim.
NOMINATIM_EMAIL empty Optional email parameter passed to Nominatim search.
NOMINATIM_MIN_INTERVAL_SECONDS 1.0 Local delay between Nominatim requests.
NOMINATIM_COUNTRYCODES ru Default country filter for geocoding.
DEFAULT_RADIUS 50000 Fallback radius in meters for coordinate searches.
KUDAGO_LOCATIONS_CACHE_TTL 900 In-memory KudaGo locations cache lifetime in seconds; use 0 to disable.
TRUST_ENV true Whether HTTP clients trust proxy environment variables.
LOG_DIR logging Directory for log files.
DEBUG 0 Enables debug logging when truthy.
MCP_TRANSPORT stdio stdio or http.
MCP_HOST 127.0.0.1 HTTP host.
MCP_PORT 8010 HTTP port.
MCP_PATH /mcp/ Streamable HTTP path.

Limitations

  • The default setup uses the public Nominatim service and is intentionally throttled to one geocoding request per second. See the public Nominatim Usage Policy.
  • For production, high traffic or strict availability requirements, use a self-hosted Nominatim instance or another geocoding provider behind a compatible endpoint.
  • Coordinate search is available only for KudaGo endpoints that support lat, lon and radius; several KudaGo endpoints require a native location slug.
  • Movie showings depend on KudaGo coverage for a city, cinema and date range.
  • Event availability, categories, tags and prices depend on KudaGo data quality.
  • This server does not buy tickets, reserve seats or verify live ticket availability.

Roadmap

  • Self-hosted geocoding deployment examples.
  • Optional provider adapters for other geocoding APIs.
  • Additional leisure data sources.
  • Ticket availability signals where a reliable data source exists.
  • Route planning between events.
  • Travel time and cost estimation.
  • Response normalization layer on top of raw KudaGo objects.

Notes for Production Use

Treat the current server as an MVP integration layer. It is a good fit for local AI-agent tooling, demos and controlled low-volume workflows. For production, replace the public Nominatim endpoint, configure a real User-Agent/contact email, add observability and decide how aggressively you want to cache geocoding results.

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