Think inside your AI world.
How do I connect BigQuery to Claude?
BigQuery's official remote MCP server means the warehouse answers questions in conversation — queries, metadata and resources, live.
Direct answer: connect the official BigQuery remote MCP server in Claude — Settings → Connectors, authorise your Google Cloud project ('claude mcp add' in Claude Code). Claude can then run queries, read metadata and inspect resources. The warehouse stops being a place you go and becomes a thing you ask.
The warehouse in the conversation
What tables exist for events, what the cohort query returns, how the schema changed — asked and answered without opening the console. Analysts keep the console for real work; everyone else finally gets answers without filing a request.
The discipline that keeps queries cheap
Worked case: your team ratified a query-cost discipline — no full-table scans on the events table outside partitioned windows, priced by one memorable month when exploratory SQL cost more than the BI tool. BigQuery bills by bytes scanned and says so; refusing unpartitioned scans is a team law about the billing, not a warehouse setting.
Queries, read under the law
Through Unl a query conversation arrives with the discipline standing: the proposed scan is unpartitioned on events — against the rule the memorable month wrote. The read is one-way; Unl runs nothing. Your cost discipline is simply present at the moment the expensive convenience gets suggested.
BigQuery to Claude puts the warehouse in conversation; BigQuery through Unl keeps your query disciplines in that conversation — the partition rule arrives before the scan does.
Reads through Unl arrive with measured context — in the presence of the decisions you’ve already settled. The reach lane is live: 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.
Read further
Questions people ask
What does the official BigQuery MCP server offer?
Queries, metadata and resource reads from your project — Google's official remote server, authorised against your cloud credentials.
Does it connect to other AI clients?
Yes; official remote MCP means ChatGPT and other clients add it through their own flows. 'bigquery mcp' is one server, many doors.
Can Unl stop an expensive query from running?
It is not an interceptor and does not sit between you and BigQuery, so it blocks nothing on its own. What it does is serve your ratified discipline, with its reasoning, into the conversation where the query is being shaped — which is where expensive scans are actually prevented, before anything is submitted. Worth being exact about the other half: a read only reads, but Unl is not incapable of running things. Where a server exposes a tool that executes, calling it takes your explicit in-turn gesture, verified at the choke point.
What if a question genuinely needs a full scan?
Run it knowing what it costs. The partition discipline is not there to forbid expensive queries — it is there so the expensive one is a decision you took with the price in view, rather than a scan you discover afterwards on the bill.
What this is
Think inside your AI world — you stay in command
Save the thoughts, decisions and targets worth keeping, each with its reasoning, carried into every AI session the moment they matter. A new unit of exchange between you and your AI: the Settled Why with standing that travels. Unprompted.
Your whole AI world. What you decided at the epicentre. It plugs into Claude, Claude Code, ChatGPT and Cursor as an MCP connector — quick to connect, in a couple of steps.
MCP native·Human settled·Model agnostic·Your data
Measured Context
Connect Unl to bring the right information into the moment.
Your sources, read against the criteria you set.
The full product, open. Free at launch.