SkillFoxx

Yandex Search API v2 MCP server over HTTP

Yandex Search API MCP server, StreamableHTTP build

A fork of Yandex's official server adding HTTP transport and the full range of search types: web, generative, image

  • 1 ★ on the repository
  • Last commit
  • Apache-2.0
MCP server

Extended access

Why this level

  • A Docker server sends queries to the paid Yandex Search API with your key
  • The tools only search and read results, including the gen_search AI answer
All reasons and checks
Russian stack

msklv/yandex-search-mcp-serverAuthor:msklv

Install

Manual install

docker run --rm -p 8766:8766 -e SEARCH_API_KEY=... [-e FOLDER_ID=...] ghcr.io/msklv/yandex-search-mcp-server:latest

A ready image from GitHub Container Registry, endpoint at http://<host>:8766/mcp.

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • A Docker server sends queries to the paid Yandex Search API with your key
  • The tools only search and read results, including the gen_search AI answer

This is third-party code. Review the repository files before installing.

What it does

The server is forked from the official yandex/yandex-search-mcp-server, which only supports stdio and always includes the YandexGPT endpoint. It adds a run_http.py file that re-exports the same FastMCP instance over StreamableHTTP at /mcp, so the container behaves like an ordinary HTTP-MCP service alongside other such services. Beyond the inherited web_search, it adds the rest of the Search API v2 endpoints: web_search_async with deferred operation polling, gen_search with an AI answer and sources, image_search and image_search_by_image. Russian region and localization are the defaults.

Who it is for. For those assembling a stack of several HTTP-MCP servers in Docker Compose who want to add official Yandex search over HTTP, not just stdio.

Good fit when

  • You need Search API v2 over HTTP in a shared Docker stack with other MCP servers
  • You want every search type at once: web, async web, generative, image-by-text and image-by-image
  • A ready image from GitHub Container Registry works for you

Not a fit when

  • You only need stdio like the original: then it is simpler to use Yandex's official server directly
  • You are not ready to pay for the Yandex Search API: the service is commercial and needs a paid key

Example request

Find images of a specific coffee machine model and give a short AI answer with sources about their reliability

MCP

Transport
http
Authentication
API key

Environment variables

SEARCH_API_KEY required, secret
Yandex Search API key, sent as Authorization: Api-Key
FOLDER_ID
Yandex Cloud folder id, required for gen_search

Limitations

This is an unofficial fork, not a Yandex product or endorsement, although some files (server.py, detail.py) are inherited from the original under Apache-2.0 with YANDEX LLC copyright. The Yandex Search API is paid; a key and usually a Yandex Cloud folder ID are required. gen_search needs the FOLDER_ID variable.

How to disable. Stop the container or server process and remove it from your MCP client configuration.

FAQ

Can it be used like the original stdio server?

Yes, python3 server.py keeps upstream behavior; run_http.py adds a separate HTTP mode.

Is FOLDER_ID required?

It is mandatory for gen_search; for the other tools it is optional and not always needed with service-account keys.

How is this server different from the official yandex-search-mcp-server?

It is a fork of the official Yandex server. The original supports only stdio and always enables the YandexGPT endpoint. The fork adds a run_http.py file that serves the same server over HTTP and provides a full set of search types: web, generative and images.

Do I need FOLDER_ID and can I use it like the original stdio server?

FOLDER_ID is set for generative search, SEARCH_API_KEY is always needed. For stdio take the Yandex original, while this fork runs in Docker: docker run -p 8766:8766. It works in Claude Desktop and Cursor.

Discussion

No questions or reviews yet.

All Yandex Search tools
Russian stack

A self-hosted MCP for Yandex Search API v2: web, image and generative search across all six Yandex indexes

MCP serverExtended access4

Yandex Search API MCP with deferred search

Yandex Search MCP Server (Async Mode)

Russian stack

A fork of Yandex's official server defaulting to deferred async search, cutting Search API cost by up to 94%

MCP serverExtended access
Russian stack

An MCP web-search and content-extraction server via Yandex and Google on top of SerpAPI, geared toward RU/CIS results

MCP serverExtended access1
Russian stack

A search MCP server on the Yandex Cloud Search API with a monitoring dashboard, Brave fallback and an offline bundle for air-gapped networks

MCP serverExtended access

README in short

The README is upfront that this is a fork of Yandex's official server and lists two changes: HTTP transport via run_http.py and support for all Search API v2 search types instead of just web_search. It gives a tool table with sync and deferred modes, environment variables, local and Docker run commands, and a web_search call example with region parameters.

README badge

Report a problem