Worked examples and use cases for the Codatum CLI (cdm) — a command-line tool for editing Codatum notebooks (.cnb.md) locally and syncing them with the server.
Each use case is a self-contained example you can clone, point at your own Codatum workspace, and run. They're designed to be driven by an AI coding agent: you describe what you want, and the agent uses cdm to explore data, write SQL, build charts, and sync your work. This repo includes agent setup for both Claude Code and Cursor, and the examples are walked through using Claude Code.
- CLI documentation: https://cli-docs.codatum.com/
- Codatum: https://codatum.com/
| # | Use case | What it shows |
|---|---|---|
| 01 | Build a dashboard through conversation | Create a notebook from scratch by talking to Claude Code — explore tables, build charts, add parameters, and lay out a dashboard. |
| 02 | Build from a template | Turn a recurring report into a reusable template, then regenerate it from a single prompt — including web research and write-up. |
| 03 | Manage with GitHub | Keep notebooks in GitHub as the source of truth — validate changes in pull requests and deploy to Codatum on merge to main. |
Each use case has its own README with the full walkthrough and a link to the corresponding documentation.
- Install and authenticate the
cdmCLI. See Getting started. - Set up your agent. This repo includes the cdm skill for both Claude Code (
.claude/skills/cdm/SKILL.md) and Cursor (.cursor/skills/cdm/SKILL.md) in the repo root. Open your agent from the repo root so it picks up the skill. The walkthroughs use Claude Code; see Claude Code setup. - Pick a use case from the table above and follow its README.
You'll need a Codatum workspace with a BigQuery connection that can query bigquery-public-data, since the examples are built on public datasets.
codatum-cli-examples/
├── .claude/skills/cdm/SKILL.md # cdm skill for Claude Code
├── .cursor/skills/cdm/SKILL.md # cdm skill for Cursor
└── use-cases/
├── 01-build-dashboard/
├── 02-build-from-template/
└── 03-manage-with-github/
The notebooks are committed with execution results stripped (cdm notebook strip) — they contain the SQL, chart definitions, parameters, and layout, but no data. Each one references a connection ID specific to the original workspace, so you'll point it at your own connection before running (each README explains how).
See LICENSE.