Think inside your AI world.

dbt through Unl

dbt's MCP server will read back a model's run status, its test results, and how fresh its sources are. Whether that health clears it as a foundation for the next dashboard is your rule to apply - a model that ran isn't the same as a model that passed.

dbt through Unl reads a model's run status, test results, and upstream freshness against the build-safety rule you have ratified, so 'is this model safe to build on?' comes back safe-or-hold with the condition it hasn't met.

What dbt holds

Ask dbt's MCP server about a model and it surfaces:

  • model health: run status, test results, and upstream source freshness
  • model details, including compiled SQL and columns
  • source freshness status across the project
  • the full lineage graph of ancestors and descendants
  • defined metrics and their compiled SQL via the Semantic Layer

What the naked read gives you

dbt's MCP server reads back a model's run status, its test results, its upstream source freshness, its full lineage, and the compiled SQL behind it. It is a detailed account of where a model stands in the DAG right now, offered without any view on whether that standing is solid enough to build the next thing on top of it.

The frame judges the data it is given; it does not verify the source’s accuracy.

What changes when dbt is read measured

An analytics engineer at a logistics marketplace ratified a build-safety rule in Unl: no new dashboard gets built on a model unless its latest run passed every test and its upstream sources are no more than twenty-four hours stale, because a dashboard built on a quietly failing model once reported delivery numbers that were wrong for a week before anyone raised it.

'Is this model safe to build the new dashboard on?' returns, measured: not yet - the model health read shows a failed uniqueness test on order_id in fct_orders, and its upstream stripe_payments source is thirty hours stale against the twenty-four-hour window. dbt supplied the health signals; your ratified rule supplied the safety bar.

And back again

Once the test passes and the source refreshes, dbt through Unl picks up the cleared health signals next time it checks; the ratified build-safety rule marks the model fit to build on, updating the one page.

The answer comes back measured against what you already decided, and why.

The lane is live and open to this tool today: one box, paste anything. If it speaks MCP, Unl can reach it. Readings arrive unprompted, the data beside the criterion; Unl is a courier, not a warehouse, and keeps only your keys and the frame.

Questions people ask

What needs to happen for this model to pass?

The uniqueness test passing on the next run, and the upstream source refreshing inside twenty-four hours. The same ratified rule reads the cleared health signals and marks the model safe to build on.

How do I connect dbt to Claude?

Install dbt Labs' official MCP server (dbt-labs/dbt-mcp) locally via uvx, or use the remote server hosted on dbt Cloud, with your project credentials either way - Unl's reads run through that same server.

Does Unl write to my dbt project?

No. Unl calls the Discovery and Semantic Layer read tools - model health, lineage, compiled SQL - and does not trigger runs or edit models. Any fix to a failing test, you make yourself.

What this is

Think inside your AI world — you stay in command

Unlimitless (Unl to friends) holds what you've settled, reads what your tools are showing, and catches what's changed out in the world — and hands your AI whatever bears on the work, the moment it's needed, without you asking. The right thing, in front of the model, unprompted, with you in command of the call. So you keep moving toward what you set out to build, on top of everything you've already decided.

It plugs into Claude, Claude Code, ChatGPT and Cursor as an MCP connector. Quick to connect, in a couple of steps.

Unlimitless is open now to invited Alpha. Apply for the Beta waitlist to come in ahead of the full launch:

Alpha is invite-only · free at launch.