yandex-tracker-mcp

A Model Context Protocol server that gives AI assistants authenticated access to Yandex Tracker issues, queues, comments, worklogs, and search, with OAuth2 support and a Russian docs version.

This is a Model Context Protocol server that lets AI assistants talk to the Yandex Tracker APIs with authenticated access. The README is also available in Russian, pointing at the alternate language version, which says something about who the project expects to use it. The access surface covers issues, queues, comments, worklogs, and search.

Features include complete queue management: listing and accessing every available queue with pagination support, tag retrieval, and detailed metadata. Pagination matters for large trackers, where the queue list gets long. User management retrieves account information, including login details, email addresses, and license status. Together they let an assistant work with tracker data through the MCP interface.

Organization configuration uses separate environment variables for Yandex Cloud organizations and Yandex 360 organizations. Which variable you set determines where the tracker API calls are scoped, so the config maps directly onto the two kinds of Yandex organization.

In Claude Desktop, installation means downloading the mcpb file from GitHub Releases, double clicking to install it, providing the Yandex Tracker OAuth token when prompted, and making sure the extension is enabled. Manual setup supports uvx or a Docker image, with authentication via tracker token, IAM token, or service account credentials, plus configuration file paths for macOS and Windows.

Queue management, user lookups, and organization scoping all hang off OAuth2 authentication. Claude Desktop gets a one click install from a downloaded mcpb file, while uvx and Docker cover the manual route.

• Source 1 · Project repository
• Source 2 · README