dbt + Metabase integration.

analytics, business-intelligence, data, data-modelling, dbt, elt, metabase, pypa, python, vizualisation
pip install dbt-metabase==1.3.1



GitHub Actions PyPI Downloads License: MIT Code style: black

Integration between dbt and Metabase.

If dbt is your source of truth for database schemas and you use Metabase as your analytics tool, dbt-metabase can propagate table relationships, model and column descriptions and semantic types (e.g. currency, category, URL) to your Metabase data model, and extract questions and dashboards as exposures in your dbt project.


Requires Python 3.8 or above.


You can install dbt-metabase from PyPI:

pip install dbt-metabase

Sections below demonstrate basic usage examples, for all CLI options:

dbt-metabase --help


Before running dbt-metabase, you need a compiled manifest.json file to parse. These are part of the dbt artifact generated during compilation.

Once dbt compile finishes, manifest.json can be found in the target/ directory of your dbt project.

See dbt documentation for more information.

Metabase API

All commands require authentication against the Metabase API using one of these methods:

  • API key (--metabase-api-key)
    • Strongly recommended for automation, see documentation (Metabase 49 or later).
  • Username and password (--metabase-username / --metabase-password)
    • Fallback for older versions of Metabase and smaller instances.

Exporting Models

Let's start by defining a short sample schema.yml as below.

  - name: stg_users
    description: User records.
      - name: id
        description: Primary key.
          - not_null
          - unique

      - name: email
        description: User's email address.

      - name: group_id
        description: Foreign key to user group.
          - not_null
          - relationships:
              to: ref('groups')
              field: id

  - name: stg_groups
    description: User groups.
      - name: id
        description: Primary key.
          - not_null
          - unique

      - name: name
        description: Group name.

This is already enough to propagate the primary keys, foreign keys and descriptions to Metabase:

dbt-metabase models \
    --manifest-path target/manifest.json \
    --metabase-url https://metabase.example.com \
    --metabase-database business \
    --include-schemas public

Open Metabase and go to Settings > Admin Settings > Table Metadata, you will notice that id column in stg_users is now marked as "Entity Key" and group_id is a "Foreign Key" pointing to id in stg_groups.

Try running dbt-metabase models --help to see all the options available for fine tuning.

Foreign Keys

Native relationship tests and column-level constraints are the recommended ways of defining foreign keys, however you can override them with fk_target_table and fk_target_field meta fields. If both are set for a column, meta fields take precedence.

- name: country_id
  description: FK to User's country in the dim_countries table.
    metabase.fk_target_table: analytics_dims.dim_countries
    metabase.fk_target_field: id

You can provide fk_target_table as schema_name.table_name or just table_name to use the current schema. If your model has an alias, provide that alias rather than the original name.

Semantic Types

Now that we have foreign keys configured, let's tell Metabase that email column contains email addresses:

- name: email
  description: User's email address.
    metabase.semantic_type: type/Email

Once you run dbt-metabase models again, you will notice that email column is now marked as "Email".

Below are common semantic types (formerly known as special types) accepted by Metabase:

  • type/PK
  • type/FK
  • type/Number
  • type/Currency
  • type/Category
  • type/Title
  • type/Description
  • type/City
  • type/State
  • type/ZipCode
  • type/Country
  • type/Latitude
  • type/Longitude
  • type/Email
  • type/URL
  • type/ImageURL
  • type/SerializedJSON
  • type/CreationTimestamp

See Metabase documentation for a more complete list.

Visibility Types

You can optionally specify visibility for tables and columns, this controls whether they are displayed in Metabase.

Here is how you would hide that email column:

- name: email
  description: User's email address.
    metabase.semantic_type: type/Email
    metabase.visibility_type: sensitive

Below are the visibility types supported for columns:

  • normal (default) - This field will be displayed normally in tables and charts.
  • details-only - This field will only be displayed when viewing the details of a single record.
  • sensitive - This field won't be visible or selectable in questions created with the GUI interfaces.

Tables support the following:

  • No value for visible (default)
  • hidden
  • technical
  • cruft

If you notice any changes to these, please submit a pull request with an update.

Other Meta Fields

In addition to foreign keys, semantic types and visibility types, Metabase also accepts the following meta fields:

- name: model_name
    metabase.display_name: another_model_name
    metabase.visibility_type: normal
    metabase.points_of_interest: Relevant records.
    metabase.caveats: Sensitive information about users.
    - name: column_name
        metabase.display_name: another_column_name
        metabase.visibility_type: sensitive
        metabase.semantic_type: type/Number
        metabase.has_field_values: list
        metabase.coercion_strategy: keyword
        metabase.number_style: decimal

See Metabase documentation for details and accepted values.


By default, dbt-metabase waits for tables and columns to be synchronized between your dbt project and Metabase database, otherwise the export fails when the sync timeout expires.

If you have known discrepancies between dbt and Metabase and wish to proceed without synchronization, set the sync timeout to zero (e.g. --sync-timeout 0). This is discouraged, because you will still encounter errors if you have a table or column in your dbt project that is missing from Metabase and dbt-metabase attempts to export it.

Exposure Extraction

dbt-metabase allows you to extract questions and dashboards from Metabase as dbt exposures in your project:

dbt-metabase exposures \
    --manifest-path ./target/manifest.json \
    --metabase-url https://metabase.example.com \
    --output-path models/ \
    --exclude-collections "temp*"

Once the execution completes, check your output path for exposures files containing descriptions, creator details and links for Metabase questions and dashboards:

  - name: number_of_orders_over_time
    description: '
      ### Visualization: Line

      A line chart depicting how order volume changes over time

      #### Metadata

      Metabase Id: __8__

      Created On: __2021-07-21T08:01:38.016244Z__'
    type: analysis
    url: https://metabase.example.com/card/8
    maturity: medium
      name: Indiana Jones
      email: indiana@example.com
      - ref('orders')

Native query questions will have SQL code blocks inside the descriptions, formatted to look nice in dbt docs. These YAML files can be committed to source control to understand how exposures change over time.

Try running dbt-metabase exposures --help to see all the options available for fine tuning.


There are 3 levels of configuration in decreasing order of precedence:

  • CLI arguments, e.g. --manifest-path target/manifest.json
  • Environment variables, e.g. MANIFEST_PATH=target/manifest.json
  • Configuration file, e.g. manifest_path: target/manifest.json

Try running --help for any command to see the full list of CLI arguments and environment variables.

A configuration file can be created in ~/.dbt-metabase/config.yml for dbt-metabase to pick it up automatically or anywhere else by specifying dbt-metabase --config-path path/to/config.yml (must come before the command). Here is an example YAML file:

    manifest_path: target/manifest.json
    metabase_url: https://metabase.example.com
    # Configuration specific to models command
      metabase_database: business
    # Configuration specific to exposures command
      output_path: models

Note that common configurations are in the outer block and command-specific ones are in separate blocks.

Programmatic API

Alternatively, you can invoke dbt-metabase programmatically. Below is the equivalent of CLI examples:

from dbtmetabase import DbtMetabase, Filter

# Initializing instance
c = DbtMetabase(

# Exporting models

# Extracting exposures

See function header comments for information about other parameters.