Guide · Agent skill
How to use the dbt Transformation Patterns agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The dbt Transformation Patterns skill gives an AI agent a set of conventions for building dbt projects: how to lay out models in staging, intermediate and marts layers, how to name them, and which tests, documentation and incremental settings to reach for. dbt (data build tool) turns raw database tables into clean analytics tables using SQL. This guide explains what the skill does, how to install it, how to ask for it, and what you need before it is useful. You do not need prior experience with agent skills to follow it.
dbt Transformation Patternswshobson/agents · MIT Download the skill Quick start
- 1Download the ZIP. Download the skill ZIP from this page.
- 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for personal use, or into .claude/skills/ in your project.
- 3Or upload it in the apps. In the Claude apps, open Settings, then Capabilities, then Skills, and upload the ZIP.
- 4Confirm dbt is set up. Make sure dbt is installed and a warehouse connection is in your dbt profiles if you want the agent to run models.
- 5Describe your transformation task. Ask in plain terms, for example to organize models into staging, intermediate and marts layers.
- 6Review the output. Check the proposed model names, layers and tests before applying them to your project.
What the skill makes the agent do
Once the skill is loaded, the agent follows dbt best-practice patterns when you ask it to build or reorganize transformations. It organizes models into layers: sources (raw data definitions), staging (one-to-one with a source, light cleaning), intermediate (business logic, joins and aggregations) and marts (final analytics tables).
It applies naming prefixes: stg_ for staging, int_ for intermediate, and dim_ or fct_ for marts. It also suggests data quality tests such as not null, unique and relationships, suggests column and model documentation, and recommends incremental models for large tables. The skill works from written guidance and example project layouts; it does not run anything by itself.
Install the dbt Transformation Patterns skill
Download the ZIP for this skill from this page. The ZIP contains SKILL.md and references/details.md. Unzip it into one of these locations:
- Personal skills folder: ~/.claude/skills/
- Project skills folder: .claude/skills/ inside your dbt project
Or upload it in the Claude apps
If you use the Claude apps rather than a local folder, open Settings, go to Capabilities, then Skills, and upload the ZIP there.
How to trigger the skill with natural requests
You do not need a special command. Describe the work in plain terms that match the skill's description, such as building data transformations, creating data models, or following analytics engineering best practices. The agent should load the skill when your request fits.
- Ask it to organize a dbt project into staging, intermediate and marts layers.
- Ask it to name new models following the stg_, int_, dim_ and fct_ convention.
- Ask it to add tests and documentation to an existing set of models.
- Ask it to make a large model incremental.
Worked example: a starting project layout
A typical request is to set up a project from scratch. The skill's example dbt_project.yml sets staging models as views in a staging schema, intermediate models as ephemeral, and marts as tables in an analytics schema. The example layout groups staging models by source system, for example models/staging/stripe/ with a stg_stripe__payments.sql file and matching _stripe__sources.yml and _stripe__models.yml files.
The source definition pattern declares raw tables with loaded_at_field and freshness thresholds (warn after 12 hours, error after 24 hours in the example), and attaches unique and not_null tests to key columns.
models/
├── staging/
│ ├── stripe/
│ │ ├── _stripe__sources.yml
│ │ ├── _stripe__models.yml
│ │ ├── stg_stripe__customers.sql
│ │ └── stg_stripe__payments.sql
├── intermediate/
│ └── finance/
│ └── int_payments_pivoted.sql
└── marts/
├── core/
│ ├── dim_customers.sql
│ └── fct_orders.sqlWhat you need before you start
The skill is guidance, so it is only useful alongside a working dbt setup. If you ask the agent to run or test a project, you need dbt (data build tool) installed. If you ask it to run models, you also need a data warehouse connection configured in your dbt profiles.
The skill runs no scripts and uses no network access, according to its listing.
Safety notes
The skill itself contains no scripts and makes no network calls. Any database access comes from your own dbt profile and the commands the agent runs in your environment, so review commands before approving them, especially any that run models against production. The skill's own guidance says not to test in production and to use a dev target.
Tips for better results
Give the agent your existing folder structure and naming so it adapts rather than rewrites. Ask it to extend the example conventions where your project already differs. For detailed patterns, the skill points to references/details.md, which contains worked YAML examples.
- Use the staging layer so cleaning happens once.
- Avoid hardcoded dates; use a variable such as {{ var('start_date') }}.
- Extract repeated logic into macros.
- Keep the dbt project in Git.
What people use it for
- Restructuring a messy dbt project into staging, intermediate and marts folders.
- Naming new models consistently with stg_, int_, dim_ and fct_ prefixes.
- Adding not_null and unique tests and column descriptions to existing models.
- Deciding which large tables should be incremental models.
- Setting up source definitions with freshness checks for a raw data feed.
Frequently asked questions
Is the dbt Transformation Patterns skill free?
The skill is published under the MIT licence. Check the licence file in the source repository for the full terms before reuse.
How do I install an agent skill?
Download the ZIP and unzip it into ~/.claude/skills/ for personal use or .claude/skills/ inside a project. In the Claude apps, you can instead upload the ZIP under Settings, Capabilities, Skills.
Does this skill need dbt installed?
Only if you ask the agent to run or test a project. The skill itself is guidance, and running models also requires a data warehouse connection in your dbt profiles.
Does the skill run code or access the internet?
According to its listing, it runs no scripts and uses no network access.
Can I use this skill with a tool other than dbt?
Not really. The naming and layering conventions are dbt-specific, so they would not apply to a different transformation tool.
Is it suitable for beginners?
It suits people new to dbt who want a sensible starting structure, but you still need a working dbt project and database to get value from its output.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use this skill if you are setting up or tidying a dbt project and want consistent layers, names, tests and incremental choices. Skip it if you do not already use dbt and a data warehouse, or if your team uses a different transformation tool.
Download the skill