Connect OpenClaw to Google BigQuery
Let an OpenClaw agent explore datasets, run SQL and load rows in Google BigQuery, with no credentials handed to the agent.
Google BigQuery is Google Cloud’s data warehouse: a place to keep event logs, product analytics, billing exports and whatever else your pipelines produce, and query all of it with SQL. Data teams build on it, but the people who need answers from it are usually not the ones who know the schema. So questions queue up — “how many active accounts did we have in August?” — and someone with warehouse access spends their afternoon writing one-off queries and pasting results into a doc.
With Claworc’s External Connections, an OpenClaw agent can work
inside your BigQuery project on your behalf. You sign in to Google once through a popup,
Composio holds the resulting token, and the agent gets an auto-generated
claworc-googlebigquery skill that tells it which BigQuery tools exist and how to call
them. The agent never sees a password, an OAuth token or an API key.
Tools available to OpenClaw agents
Section titled “Tools available to OpenClaw agents”Composio’s BigQuery toolkit exposes 63 tools, covering datasets, tables, jobs, routines, reservations and Analytics Hub. These are the ones an OpenClaw agent reaches for most often:
- Discover datasets and tables — see what exists in the project, so the agent can orient itself before writing any SQL.
- Read table schemas — get a table’s columns and types without scanning rows, which keeps exploratory queries cheap and correct.
- Run SQL queries — run a query and get the rows back in one call. This is what answers most questions.
- Run long jobs — start a long-running query, load, extract or copy job and collect the results when it finishes.
- Create tables and insert rows — create a table and stream rows into it, for when the agent is producing data rather than reading it.
- Monitor jobs — see what has been running in the project, useful for spotting a runaway query or checking whether last night’s load finished.
Every tool is called through a small broker running next to the agent, so the agent only needs to describe what it wants done in plain language and let the skill handle the rest.
Use cases
Section titled “Use cases”Answer questions in plain language
Section titled “Answer questions in plain language”“How many orders over $500 did we ship to Germany last quarter?” The agent lists the relevant tables, reads their schemas, writes the query, runs it and returns the number with the SQL it used, so anyone can check the logic. The data team stops being a ticket queue for simple questions.
Scheduled reports without a BI project
Section titled “Scheduled reports without a BI project”Ask the agent to run the same handful of queries every morning and post a short summary: yesterday’s signups, revenue, error rate. Because it has the schema, it can also explain an unexpected number by breaking it down one level further before you ask.
Load results from other systems
Section titled “Load results from other systems”An agent that already has a Fathom, Gmail or CRM connection can pull records from those services, create a table and stream the rows in. That gets small but useful datasets into the warehouse without waiting for a pipeline to be built.
Get started
Section titled “Get started”Connecting BigQuery takes three steps: an admin adds a Composio API key to Claworc’s
settings, you open the instance’s Settings tab, click + Add connection, pick
Google BigQuery and sign in to Google. The claworc-googlebigquery skill appears in the
instance a moment later. The full walkthrough, including the exact Composio permissions
the key needs, is in the External Connections documentation.