DataLens MCP connector

MCP server for Claude that exposes Yandex DataLens operations as tools. It follows the FastMCP + Streamable HTTP deployment pattern from the Habr article: endpoint /mcp, tool annotations, TCP healthcheck, and nginx buffering disabled for streaming responses.

What it can do

• Check DataLens API connectivity.

• Run controlled raw API requests for unsupported/new DataLens endpoints.

• List workbook or collection entries.

• Create a dashboard from a JSON payload.

• Build a simple dashboard draft spec from a compact chart/layout description.

The connector intentionally keeps DataLens payloads transparent. DataLens dashboard/chart schemas can change, so Claude can generate or adjust JSON payloads while the MCP server handles authentication, HTTP calls, validation, and deployment transport.

Related MCP server: Databricks MCP Server

Setup

python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -e ".[dev]"
Copy-Item .env.example .env

Edit .env:

DATALENS_TOKEN=your_token
DATALENS_API_BASE=https://datalens.yandex.cloud/api
DATALENS_AUTH_SCHEME=Bearer

Run locally:

datalens-mcp

The MCP endpoint is:

http://localhost:8000/mcp

Claude

For a remote connector, deploy it behind HTTPS and add this URL in Claude:

https://your-domain.example/mcp

In Claude: Settings -> Connectors -> Add custom connector.

Docker

docker compose up -d --build

The compose file binds the container to 127.0.0.1:8096, so expose it through nginx with TLS. See nginx.example.conf.

MCP tools

• datalens_healthcheck

• datalens_raw_request

• datalens_list_entries

• datalens_get_entry

• datalens_create_dashboard

• datalens_build_dashboard_spec

Notes

The official DataLens documentation has a DataLens Public API section. If a concrete endpoint differs in your account/API version, update DATALENS_API_BASE, DATALENS_ENTRIES_PATH, DATALENS_DASHBOARDS_PATH, or call datalens_raw_request with the exact path.