DataLens Dev MCP
datalens-dev-mcp
Local MCP server and Codex plugin for Yandex DataLens dashboard development: read, author, publish, back up and clean up objects
Install
python3 -m venv .venv
.venv/bin/python -m pip install .
.venv/bin/datalens-dev-mcp --versionRun inside a clone of github.com/ADIKANT/datalens-dev-mcp. The stdio connection is described in .mcp.json.
This is third-party code. Review the repository files before installing.
What it does
The server gives the agent typed operations on DataLens objects through the official datalens-sdk. The agent reads workbooks, datasets, charts and dashboards, builds drafts from eight ready visualization recipes, validates them and saves with a mandatory readback. Publishing is a separate step from a fresh saved revision only. Maintenance covers snapshot export and dependency-safe cleanup. Five bundled skills route the agent through dashboards, datasets, Editor, inspection and maintenance.
Who it is for. BI analysts and developers who build and maintain Yandex DataLens dashboards.
Good fit when
- You need to audit a dashboard without changing it and find broken relations
- You need a chart built from a dataset using a standard recipe
- You need to update a dashboard without breaking layout and selectors
Not a fit when
- You need to change DataLens access rights or licenses: not supported
- You lack rights to the organization's objects or cannot give the agent an IAM token
Example request
Audit this DataLens dashboard without changes and list broken selector relationsLimitations
Not a Yandex product; the server acts with the current user's rights. Requires Python 3.11, DATALENS_ORG_ID and DATALENS_IAM_TOKEN. The project is alpha with no PyPI package; install from a cloned repo. Snapshot export does not guarantee full restore, and license revocation and ACL changes are not supported. Browser checks are read-only and do not replace data checks.
How to disable. Remove the server from .mcp.json or the client config, and disable the datalens-dev-mcp plugin in Codex. Credentials live in ~/.config/datalens-dev-mcp/credentials.env.
MCP
- Transport
- stdio
- Authentication
- API key
| Environment variables | |
|---|---|
| DATALENS_ORG_ID required | DataLens organization ID |
| DATALENS_IAM_TOKEN required, secret | Yandex Cloud user IAM token |
| DATALENS_ENV_FILE | Path to a credentials file; takes priority |
Security check
- Creates, publishes and deletes objects in a production DataLens organization
- Uses a Yandex Cloud user IAM token
README in short
The Russian and English README covers quick start in a Python virtual environment and credential setup. It lays out the workflow: inspection, datasets and Wizard, Editor, visualization recipes, delivery with saved readback and maintenance. It lists honest limitations and official SDK boundaries. Code is Apache-2.0 with some materials under CC-BY-4.0.
FAQ
Does the server call a language model?
No. It has no built-in model, arbitrary RPC or browser writes, only typed operations.
What if a write response is lost?
The write is not retried; dl_operation_reconcile checks the result by known ID or revision.
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 token-lean MCP server for PostgreSQL, MySQL, MariaDB, SQL Server and SQLite with a read-only mode