yandex-tracker

Yandex Tracker MCP Server with OAuth2 support

Installation

Run this command in your terminal to add the MCP server to Claude Code.

Run in terminal:

Command

claude mcp add --transport stdio aikts-yandex-tracker-mcp uvx yandex-tracker-mcp@latest \
  --env TRACKER_TOKEN="your_tracker_token_here" \
  --env TRACKER_ORG_ID="your_org_id_here" \
  --env TRACKER_CLOUD_ORG_ID="your_cloud_org_id_here"

How to use

The Yandex Tracker MCP Server provides a secure, authenticated bridge that lets AI assistants interact with Yandex Tracker APIs. It exposes functionality for managing queues, issues, users, worklogs, and search queries through a Model Context Protocol (MCP) interface. The server also supports optional Redis caching to improve performance and offers multiple transport options (stdio, HTTP, and remnants of SSE) to integrate with various clients. By configuring the proper environment variables and starting the MCP, you enable your AI agents to query and mutate Yandex Tracker data in a controlled and scalable manner.

Once running, you can connect your MCP client (Claude Desktop/Code, Windsurf, Cursor, or any MCP-enabled frontend) using the documented configuration examples. The available capabilities include listing queues, retrieving and creating issues with custom fields, handling attachments, performing workflow transitions, and executing advanced Yandex Tracker queries. OAuth 2.0 authentication support provides token-based access with automatic refresh as an alternative to static API tokens.

To use the server effectively, supply your Yandex Tracker credentials and organization identifiers via environment variables. The server will then proxy requests to Yandex Tracker endpoints, applying any configured access restrictions and caching rules to optimize responses.

How to install

Prerequisites:

uv (global) installed on your system
A valid Yandex Tracker API token with appropriate permissions

Installation and setup (UVX):

Install the MCP using uvx from NPM:

uvx yandex-tracker-mcp@latest

Provide required environment variables (examples):

TRACKER_TOKEN: your Yandex Tracker OAuth token
TRACKER_CLOUD_ORG_ID: your Yandex Cloud organization ID
TRACKER_ORG_ID: your Yandex 360 organization ID (if applicable)

Example usage (uvx):

TRACKER_TOKEN=your_token TRACKER_CLOUD_ORG_ID=your_cloud_id TRACKER_ORG_ID=your_org_id uvx yandex-tracker-mcp@latest

Start the MCP server via uvx (in the background or foreground as per your workflow).

Alternative: Docker If you prefer Docker, you can run the MCP image directly:

docker run --rm -i \
  -e TRACKER_TOKEN \
  -e TRACKER_CLOUD_ORG_ID \
  -e TRACKER_ORG_ID \
  ghcr.io/aikts/yandex-tracker-mcp:latest

Confirm the server is reachable from your MCP client using the provided config example.

Note: You can also use the Claude Desktop/Code integration or Windsurf/Cursor configurations shown in the repository to easily wire up the MCP server with these tools.

Additional notes

Environment variables: Always provide one of the authentication methods (TRACKER_TOKEN, TRACKER_IAM_TOKEN, or service account credentials) as documented. You can choose TRACKER_CLOUD_ORG_ID or TRACKER_ORG_ID depending on whether you’re using Yandex Cloud or Yandex 360 organizations. If you enable Redis caching, ensure your Redis instance is accessible and properly secured. The server supports multiple transports; pick stdio for local testing (via Claude/Code or Windsurf) or HTTP for broader integration. If you encounter token refresh issues, verify token scope and expiration settings in Yandex Tracker. For debugging, enable verbose logs in your MCP client and monitor container or process logs for transport-related messages.