The news
Google, part of Alphabet (GOOGL), made Data Agent Kit generally available in a September 30, 2026 post on the Google Cloud blog, giving outside AI coding agents a direct line into BigQuery and more than 15 other Google Data Cloud services.
The post, by product manager Arun Nair and developer advocate Jeff Nelson, describes the kit as a free set of Model Context Protocol tools and agent skills. Model Context Protocol, or MCP, is an open standard that lets AI agents call outside tools and data. The MCP tools let an agent inspect schemas, run queries, read job logs and manage resources, while the Google-written skills are open-source instructions on practices such as optimizing BigQuery SQL, designing Bigtable row keys and building dbt pipelines.
Google lists support for its own Antigravity tools alongside rival agents, including Anthropic's Claude Code and OpenAI's Codex, plus VS Code and compatible editors such as Cursor. The kit comes pre-installed in Cloud Shell and Cloud Workstations. Google's documentation also lists Gemini CLI among the supported agents.
Services covered include BigQuery, Spanner, AlloyDB, Cloud SQL, Bigtable, Managed Service for Apache Spark, Lakehouse for Apache Iceberg, dbt, Dataform, Managed Service for Apache Airflow and Dataplex Knowledge Catalog. New at general availability, according to Google, are BigQuery Graph, Bigtable support, stateful serverless Spark sessions, and federated queries into AWS Glue and Databricks Unity Catalog without building ingestion pipelines.
Google said the kit costs nothing extra; customers pay standard prices for the Google Cloud services their agents use. On security, Google said agents run as the user or an impersonated service account, so row- and column-level security applies automatically, and administrators can govern access with IAM policies, screen MCP traffic with Model Armor, and fence agents in with VPC Service Controls.
The numbers
- Google Data Cloud services connected
- 15+
- Extra cost for the kit
- $0 (standard service pricing applies)
- GA blog post date
- September 30, 2026
Why CEOs should care
For CIOs and heads of data, the practical point is that developers no longer need to use Google's own assistant to work deeply with Google's data stack. If your engineers already use Claude Code, Codex or Cursor, they can now point those agents at BigQuery and Spanner with Google-supported tooling. Ask your data team which agents they use today and whether standardizing on this kit would replace home-grown connectors.
CISOs should treat an agent with database access as a new kind of user. Google's design runs agents under a person's identity or a service account, which keeps existing permissions in force, but it also means an over-privileged account becomes an over-privileged agent. Review service account scopes, decide whether to turn on Model Armor screening and VPC Service Controls, and make sure query activity from agents is logged and attributable.
CFOs should note that while the kit is free, agent-written queries are not. BigQuery and Spark bill on use, and an agent can run many queries quickly. Google says its skills nudge agents toward cost-aware habits such as dry runs and partition checks; finance teams should still set budgets and alerts on projects where agents operate.
The bigger picture
The release is a bet that developers will keep switching AI coding tools, and that the cloud which works best with all of them keeps the data workloads. By supporting rivals' agents rather than only its own, Google aims to make BigQuery the default destination whichever assistant wins a team's favor. The federation features, which reach into AWS Glue and Databricks catalogs, extend the same logic to data that lives outside Google Cloud.
What’s next
Watch whether Amazon Web Services, Microsoft and Snowflake ship comparable agent-neutral toolkits for their data platforms, and whether Google adds more services beyond the 15-plus covered at general availability.
What “Fact-checked” means
Fact-checking means testing a story’s facts against the evidence before it is published. This story went through at least two separate checks before this version was published.
- What we checked
- Its names, figures, dates, job titles, quotes and who said what were checked against the story’s sources, including its main source where it could be opened. The headline was checked for accuracy and overstatement.
- How
- A first check reviewed the whole story. If it passed, a second, skeptical check went back to the sources to look for mistakes in the most important facts. If a check flagged the story, it was edited to fix the problems found, and a separate re-check then reviewed the whole story again.
- Who
- The checks are made by our newsroom, as steps kept separate from the writing, under rules set by our editor, Hussein Mukhtar. A story the checks still flag is held for the editor, who decides whether it is fixed, published or dropped.
- If something is wrong
- “Fact-checked” does not mean error-free. If a material error is found after publication, we correct the story and add a note saying what changed. Report an error
Companies in this story








