Data & Analytics · dbt Cloud + Claude
dbt Cloud MCP with Claude: Query Governed Data in Plain Language
The short answer
Claude works with dbt Cloud through the Model Context Protocol (MCP), an open standard that lets Claude read and act within the tool using permissions you control. Once connected, Claude can inspect projects, discover metadata, monitor job runs and query governed data through natural language, with access limited to what your team approves.
What is dbt Cloud?
dbt Cloud is a platform for managing analytics projects and governed data products. It provides administrative APIs, metadata discovery and Semantic Layer access, so teams can build, test and document data transformations in one place. Data engineers, analytics engineers, data analysts and data leaders use it to keep business metrics consistent and trustworthy across the organisation.
How Claude connects to dbt Cloud through MCP
The Model Context Protocol is an open standard that works like a universal adapter between Claude and other software. It lets Claude use a tool securely, reading information and taking actions, with permissions the organisation controls. For dbt Cloud, the connection can use the tool's official MCP server or a direct API route, depending on what best fits your setup.
Once connected, Claude can typically look up projects and their structure, search metadata about models and definitions, check the status of scheduled and triggered job runs, and ask the Semantic Layer for governed metrics. A business user can ask a question in plain English and receive an answer based on the official definitions your team has already agreed.
Permissions and approvals keep people in charge. Access is granted with least privilege, using either an access token or OAuth sign-in as your organisation prefers. Actions that change jobs or projects can be routed for human approval, and every interaction can be logged for audit and review.
What you can automate
Answering metric questions in plain English
A sales manager asks Claude for last quarter's net revenue by region. Claude queries the governed metric in the Semantic Layer, so the answer uses the same definition as the official dashboards. This reduces conflicting numbers and saves analysts from handling routine requests.
Finding where a data definition lives
A new analyst wants to know which model feeds the customer churn figure. Claude searches the project metadata and explains the lineage and documentation. Onboarding becomes faster because people can explore the project without waiting for a senior colleague to explain it.
Monitoring job runs and spotting failures
A data operations lead asks Claude which scheduled jobs failed overnight. Claude checks recent job run statuses and summarises the failures by project. The team can respond early, before business reports are affected, and can focus on the jobs that need attention first.
Preparing project status updates
A head of analytics needs a weekly summary of data product health. Claude gathers job run results and project information, then drafts a short status note. Leaders receive a clear picture without the data team spending hours compiling it by hand.
Supporting governance and documentation reviews
A data governance officer asks Claude to list models that lack descriptions or owners. Claude reviews the metadata and produces a checklist for the team to act on. Documentation gaps are found and closed in a structured way, supporting internal audit preparation.
An end-to-end workflow: Weekly revenue briefing for the leadership team
TriggerEvery Monday at 7:00 am, a scheduled request is sent to Claude from the operations workspace.
Claude checks dbt Cloud for job runs that completed over the weekend and flags any failures.
Claude queries the governed revenue and churn metrics through the Semantic Layer for the previous week.
Claude compares the results with the prior week and identifies notable changes by region.
Claude drafts a short briefing and places it in Microsoft 365 for the analytics lead to review.
The analytics lead checks the figures, adds commentary, and approves the briefing before it is sent.
The approved briefing is logged in the CRM account notes for the sales leadership team.
OutcomeLeaders receive a consistent, reviewed weekly briefing based on agreed metric definitions, typically saving several hours of manual compilation each week while keeping a person in control of what is shared.
How we set it up
Free automation audit
We review how your data team works today and identify the workflows around dbt Cloud that are worth automating. You receive a short written summary of priorities at no cost.
Secure connection
We connect Claude to dbt Cloud using the official MCP server or a direct API route. Access is set up with either an access token or OAuth sign-in, whichever your security team prefers, with least-privilege permissions.
Workflow design
We design the end-to-end workflow across dbt Cloud and your other systems such as Microsoft 365, CRM or ticketing. People approve any consequential actions before they take effect.
Testing with real cases
We test the workflow against past questions, job runs and metric requests from your team. We refine prompts, permissions and outputs until results match what your analysts expect.
Launch, monitoring and training
We launch with monitoring and audit logs in place, then train your analysts, operations staff and leaders on how to use the workflow confidently and safely.
Who benefits
- Head of Analytics
- Consistent metric answers and fewer repeated requests reaching the data team.
- Data Engineer
- Faster visibility into job run health and quicker diagnosis of failures.
- Business Operations Manager
- Self-service answers from governed data without waiting for a report.
- Data Governance Officer
- Easier review of metadata and documentation to support audit readiness.
Security and control
- Least-privilege access, granting Claude only the projects and permissions it needs.
- Choice of access token or OAuth sign-in, aligned with your organisation's security policy.
- Human approval for any action that changes jobs, projects or shared outputs.
- Audit logs of every interaction, so you can review what was asked and answered.
- Testing against real cases before launch, with ongoing monitoring afterwards.
For teams in the US, UK, India and the Middle East
Organisations in the United States and the United Kingdom can host the solution in their chosen region to meet internal data handling and sector requirements. Companies in India must consider the Digital Personal Data Protection Act, 2023, and we design hosting accordingly. In the Middle East, including the UAE and Saudi Arabia, data protection laws shape where data may be stored and processed. We select the hosting region for each deployment to meet these requirements.
Learn more about Claude and MCP
Anthropic's AI model and assistant.
Secure links between Claude and business tools.
How engineers build and change software with Claude.
Hosted agents that run multi-step work.
The open standard Claude uses to connect to tools.
Frequently asked questions
What does the dbt Cloud MCP server do?
The dbt Cloud MCP server is a standard connection point that lets AI tools such as Claude interact with dbt Cloud in a controlled way. It can typically help inspect projects, discover metadata, monitor job runs and query governed data. It acts as a secure adapter, so Claude only uses the capabilities you permit.
Is connecting Claude to dbt Cloud secure?
Security depends on how the connection is configured. We use least-privilege access, either an access token or OAuth sign-in, and audit logs of activity. Consequential actions can require human approval. We design each connection to match your organisation's security policy before launch.
What can Claude do with dbt Cloud once connected?
Typically, Claude can look up projects, search metadata about models and definitions, check job run statuses, and query governed metrics through the Semantic Layer. Users ask questions in plain English. The exact actions allowed are set by your permissions, so you decide what Claude can read and change.
How long does setup take?
Timelines vary with the size of your data estate and the workflows you choose. A focused connection with one or two workflows can often be launched within several weeks, including testing and training. We begin with a free automation audit to give you a realistic timeline before any work starts.
Which dbt Cloud plan or permissions are needed?
The plan and permissions required depend on the features you want to use and how your account is configured. Generally, the connection needs access credentials with appropriate project permissions. During the free audit we confirm what your current setup supports and what, if anything, needs to change.