We've launched the beta version of "Codatum CLI" — Codatum's functionality, now operable from your local environment, in collaboration with AI agents.

From fetching, editing, running, and syncing notebooks to executing SQL and searching data sources, you can complete your entire analytics workflow from the command line alone, without ever opening a browser.

Notebooks can be retrieved as plain text files in Markdown format (.cnb.md), enabling a true "BI as Code" workflow: edit in your favorite editor, version control with Git, and review through Pull Requests.

Analytics work that was previously confined to the browser can now be completed entirely with just a terminal and an editor.

Key Features

Edit notebooks as text files: Use cdm notebook clone to pull a notebook to your local machine, and sync with the server via push / pull. You can check diffs anytime with cdm notebook diff.

Run and validate SQL: Execute SQL directly against a connection with cdm sql, and pre-check the validity of SQL and schemas within a notebook using cdm notebook validate / format.

Search data sources: List connections with cdm connection list and search table/column information with cdm catalog search-tables — understand what data is available without opening the GUI.

Live sync preview with the browser: cdm notebook preview bidirectionally syncs your local files with the browser, letting you edit interactively while viewing execution results and charts.

CI/CD integration: Since the same CLI commands run in CI as well, you can build pre-merge validation and rendering checks directly into your pipeline.

Designed for AI agent collaboration: Agents can reference Codatum's specifications directly via cdm doc, preventing incorrect SQL generation caused by schema assumptions — so you can hand off notebook editing to agents like Cursor and Claude Code with confidence.

Use Cases

Build dashboards through conversation
Just tell it "I want to build a dashboard from this data," and Claude Code will explore the tables, construct the SQL, and turn it into charts. Watch the browser preview and simply say "fix this part" to shape the dashboard interactively.

Turn routine work into templates
Once you've built a template, you can hand off everything — from updating data to writing up insights to pushing to the server — with the same standard prompt each time. Weekly and monthly report creation is done with a single prompt.

Manage with GitHub
Manage notebooks on GitHub, review changes via Pull Request, and merge to main — changes sync to Codatum automatically. Even with multiple contributors, you can track who changed what and grow your dashboards safely.

For installation instructions and command details, please see the support documentation. (The support documentation is written in Japanese. Feel free to have your AI read and make use of it.)
Specifications are subject to change without notice.

We hope this helps with your data analysis! We'd love to hear your feedback and feature requests as you try it out.