Working with the Public API

DataLens provides a public API, https://api.datalens.tech, to automate operations with dashboards, charts, datasets, and connections.

The DataLens Public API offers methods similar to those used in the DataLens web interface. This API is described using the OpenAPI specification and authenticated via Yandex Cloud IAM tokens.

When using the API, please be aware of the limits in place.

To use the API, you need an IAM token and an organization ID.

For example, substitute your <IAM_TOKEN>, <ORG_ID>, and <ENTRY_ID> values into the following request to get a list of all objects related to the entity:

curl -X 'POST' \
  'https://api.datalens.tech/rpc/getEntriesRelations' \
  -H 'accept: application/json' \
  -H 'x-dl-api-version: 1' \
  -H 'Authorization: Bearer <IAM_TOKEN>' \
  -H 'x-dl-org-id: <ORG_ID>' \
  -H 'Content-Type: application/json' \
  -d '{
  "entryIds": [
    "<ENTRY_ID>"
  ]
}'

MCP serverMCP server

The DataLens MCP server provides LLM agents with access to the public API.

The server uses the MCP protocol via stdio. Add it to the MCP client configuration using one of the following methods:

• Via npx (recommended). Run this request for npx to download the published package, no manual installation or compilation required:

  {
    "mcpServers": {
      "datalens": {
        "command": "npx",
        "args": ["-y", "@datalens-tech/mcp@latest"],
        "env": {
          "DATALENS_ORG_ID": "<org-id>"
        }
      }
    }
  }
  

• Using global installation. Install the package once, then use the datalens-mcp command:

  npm install -g @datalens-tech/mcp
  

  {
    "mcpServers": {
      "datalens": {
        "command": "datalens-mcp",
        "env": {
          "DATALENS_ORG_ID": "<org-id>"
        }
      }
    }
  }
  

• From a local build. Use this option if you cloned the repository. Compile it first:

  npm ci
  npm run build
  

  Then point the client to the compiled file:

  {
    "mcpServers": {
      "datalens": {
        "command": "node",
        "args": ["/absolute/path/to/datalens-mcp/dist/index.js"],
        "env": {
          "DATALENS_ORG_ID": "<org-id>"
        }
      }
    }
  }
  

Where DATALENS_ORG_ID is your organization's ID.

By default, the server runs yc iam create-token to get an IAM token.

If you do not have the Yandex Cloud CLI yet, install and initialize it.

Additional yc settings:

• DATALENS_YC_PROFILE: Use a specific yc profile instead of the active one.
• DATALENS_YC_BIN: Full path to the yc binary file if not located in PATH.

If you cannot run yc (e.g., in an isolated environment), you can manage the IAM token yourself: install DATALENS_YC_STATIC_AUTH=1 and put the token in DATALENS_API_AUTH_HEADER="Bearer <iam-token>". The value will be sent on each request, unmodified, and yc will never be called.

Environment requirementsEnvironment requirements

To run an MCP server alongside an agent or MCP client, make sure your environment meets these prerequisites:

• Node.js version 18 or higher.
• npm and npx, accessible via PATH (usually bundled with Node.js).
• yc CLI if using the recommended authorization method via Yandex Cloud. When authorizing with a yc static token, the CLI is not required.

Use casesUse cases

• Use cases on GitHub