Yandex DataLens public API MCP server
datalens-mcp
A Rust MCP server for the Yandex DataLens public API: 60 methods via a full catalog, typed wrappers for common operations and IAM auth
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- The datalens.editor or admin role can create and change datasets and connections
- The IAM token grants access across the whole Yandex Cloud organization
Install
Manual install
Get DATALENS_ORG_ID from the Yandex Cloud console and an IAM token via yc iam create-token after yc init, then install the prebuilt binary per the README for your OS and connect the server with the YC_IAM_TOKEN and DATALENS_ORG_ID variables.
This is third-party code. Review the repository files before installing.
What it does
The server is written in Rust and talks to the public DataLens API at api.datalens.tech over stdio. The datalens_list_methods tool returns the full 60-method catalog from a February 18, 2026 snapshot, datalens_get_method_schema returns the request schema and examples for a given method, and datalens_rpc makes a generic call to any catalog method, including ones added later with no typed wrapper. Separately there are ready typed tools for high-frequency operations: reading and creating connections and datasets, dataset validation, reading dashboards, charts of several kinds (wizard, editor, QL) and workbooks, plus folder navigation and entry permissions.
Who it is for. For analysts and developers working with dashboards and datasets in Yandex DataLens who want to read them through an agent.
Good fit when
- You need to pull a DataLens dashboard or dataset structure through an agent
- You need a DataLens API method with no typed wrapper yet: datalens_rpc covers it
- You need to validate a dataset before using it in a chart
Not a fit when
- You have no Yandex Cloud organization or IAM token: the DataLens Public API will not let you in without them
- The IAM token needs manual renewal if obtained via yc CLI or OAuth exchange: the server does not refresh it itself
Example request
Show the structure of this DataLens dashboard by its ID and the list of datasets in that workbookLimitations
This is explicitly an unofficial project: the README has a dedicated section disclaiming any Yandex affiliation, noting the name is used only to describe API compatibility. It needs the datalens.editor or datalens.admin role and a not always obvious path to an IAM token, via yc CLI, an OAuth exchange, or a service account with a key. IAM tokens live for a limited time, up to 12 hours when obtained via OAuth exchange.
How to disable. Remove the server from your MCP client config.
MCP
- Transport
- stdio
- Authentication
- OAuth
| Environment variables | |
|---|---|
| YC_IAM_TOKEN required, secret | Yandex Cloud IAM token, with a limited lifetime, up to 12 hours when obtained via OAuth. |
| DATALENS_ORG_ID required | Organization ID in Yandex Cloud. |
Security check
- The datalens.editor or admin role can create and change datasets and connections
- The IAM token grants access across the whole Yandex Cloud organization
README in short
The English README opens with a direct unofficial-status disclaimer, walks through install options for Linux, Fedora, Debian and Ubuntu, macOS and Windows, or building from source, lists four ways to get DataLens API credentials, and details three paths to an IAM token: via yc CLI, an OAuth exchange, and a service account with a key.
FAQ
Is this an official Yandex server?
No, the README states plainly that the project is unofficial and not affiliated with or endorsed by Yandex.
What if a method has no typed wrapper yet?
Use datalens_rpc, which calls any method from the datalens_list_methods catalog directly.
Related
166 skills for scientific work: bioinformatics, cheminformatics, clinical data, geospatial analysis and 100+ databases
Google's open-source MCP server for databases: ready tools for Postgres, MySQL, BigQuery, Spanner and more, plus custom tools in tools.yaml
Official Hugging Face skills: Hub operations via the hf CLI, datasets, model training, Spaces, evals and deployment
A visualization language for agents and an MCP server: neat charts from a simple semantic spec