Databox CLI - Your agentic analytics platform in the terminal
by•
Create data sources, load data, build metrics, and automate workflows directly from your terminal. Use 90+ commands yourself, or let AI coding agents like Claude Code, Cursor, and Codex do the work with 10 ready-to-use agent skills.
Replies
Hi Product Hunt! 👋
I’m Tadej from Databox. Today we’re launching Databox CLI, bringing Databox directly into your terminal.
For most people, Databox lives in a browser tab. You log in, check dashboards, and manage your analytics in the app. Meanwhile, teams increasingly use scripts, data pipelines, CI jobs, and AI coding agents to move data between systems, configure tools, and automate repetitive tasks.
Until now, bringing Databox into those workflows meant working directly with our API, writing and maintaining custom code, handling authentication, retries, and errors, and checking that everything worked as expected.
This is exactly the kind of friction we want to remove as part of our broader vision to open up the Databox platform, making it easier for people and AI agents to work with Databox through the tools they already use.
That's why we built Databox CLI. It brings the capabilities of our public API directly into your terminal, so you can build dashboards and metrics, manage your data, and automate repetitive tasks in Databox using simple commands.
Here’s what you can do with it:
Create datasets with defined schemas: set column types and primary keys, and validate records before they’re ingested.
Bring in data from anywhere: Load JSON data from files, scripts, or directly from pipelines. For example: curl -s https://api.internal/export | databox dataset ingest DATASET_ID
Verify every ingestion: Confirm that your data arrived successfully and see exactly what went wrong if it didn't.
Build metrics: Create metrics on top of your datasets and trace where the data comes from.
Manage your Databox account: Work with users, permissions, databoards, and connections.
That's 91 commands across 14 areas, with a 1:1 match to the Databox public API.
And you can let AI coding agents do the work for you, too.
Databox CLI comes with 10 ready-to-use agent skills for Claude Code, Cursor, Codex, and other compatible coding agents. You can ask your agent to set up a dataset, load data, verify the results, and build metrics, and it can execute those steps for you.
We've also built in safeguards for reliable automation. Commands support structured JSON output with --json, clear exit codes, and errors that explain what went wrong. Idempotency keys make retries safe, and destructive actions require an explicit --force flag.
Want to try it?
Getting started takes three commands:
npm install -g databox-cli
databox auth login
databox account list
To add the agent skills: npx skills add databox/databox-cli --skill '*'
Databox CLI is available on all Databox plans and is open source under the MIT license.: github.com/databox/databox-cli, available as a node npm package https://www.npmjs.com/package/databox-cli.
I’ll be here in the comments all day. I'd love to hear what you'd build or automate with Databox CLI. Share your use case, and I'll help you figure out the commands to make it happen. 🙏
Been looking forward to this one team - awesome work, especially for going full agentic.
My first use case with this (as someone in sales) is to use my agent to automate my outreach cadence to people from product signals I have in the database - work that is now being done manually (with great results), but can free up so much time for my AEs to work on the meetings they have.
@zan_perkovic Thanks Žan! That’s a great first use case. Product signals are exactly the kind of data an agent should act on, not just report on. If the signals live in a Databox dataset, your agent can pull the rows with dataset data and build the outreach list from there. Would love to hear how it goes once it’s running.
@zigapotocnik will update you for sure!
Good to see a proper CLI with the agent skills already in the box. I've ended up giving my agents command line tools against the API for most things, so this is pretty much how I'd want to hook an agent up to analytics. Did the 91 commands come straight from the API spec?
@simon_moxon Thanks Simon! Yes, every command maps 1:1 to an endpoint in the Databox public API (V2), so there’s no extra logic between your agent and the data. The 10 agent skills are split by area (datasets, metrics, data sources, users, etc.), so an agent only loads the part it needs. Agents also get --json, clear exit codes, and --idempotency-key for safe retries on writes. Curious what you’ve been using for your other API tools - anything you’d want us to add?
Awesome work, team! 11th launch in the 11th month. I'm not a big fan of angel numbers, but let this be the lucky one for you ;)
@fedulovelena Thanks Helen! We’ll take all the luck we can get. Fingers crossed the numbers are on our side today. 🍀
Congrats on the launch!
@neo_tiangratanakul Thanks Neo, appreciate the support!
Nice having all the skills already built into the CLI. Best of luck with the launch today.
@codeandsea Thanks Brent! That was the goal - install the CLI, add the skills, and your agent is ready to work with your Databox data. No extra setup or wiring needed. Appreciate the support today!
I tested it on a finance monthly report we update every month from a CSV export and Excel. I asked Claude Code to set up a dataset with the right columns and load the file. With skills installed, it picked the commands, defined the schema & checked that the import went through. Next month it's the same command with a new file. Works well, great!
@tadej_kelc Love this, Tadej - that’s the workflow we had in mind. Set it up once, then it’s one command each month. One tip for recurring files: set a primary key on the dataset (--primary-key). Then if a month’s file gets re-sent or corrected, matching rows get updated instead of duplicated.
We kept things simple and predicable.
Every CLI command maps directly to the @Databox public API. No hidden magic or extra layers of abstraction to figure out.
I think this becomes especially important when you start letting AI agents execute commands on your behalf. You want to understand exactly what is happening behind the scenes.
And since its fully open source under MIT, you can check the implementation yourself 🚀
We built this because we kept seeing agents get stuck at the same point. They could write the code, but they couldn't touch the Databox setup. With the skills, an agent can go from 'track this' to a data source, a dataset with the right schema, and data flowing in, all in one session.
Hey Product Hunt 👋
AI agents are great at writing code. Things get interesting when you give them the tools to do the work.
Databox CLI is that toolset for analytics. Tell your agent “set up our orders data, load last quarter, and build the revenue metrics.” It does the work, checks it, and reports back. You don’t write any custom API integration.
What’s inside:
91 commands, a 1:1 mirror of our public API
10 ready-to-use skills for Claude Code, Cursor, Codex and other coding agents
Schema validation on every record and explicit confirmation for destructive commands, because you’re trusting an agent with real data
This is the third step in opening Databox up. First came the public API, then MCP, and now the CLI.
My question for you: if your agent could run your analytics, what’s the first thing you’d hand over?