Inspect Metabase metadata with AI
Explore database, table and field definitions from a local Metabase metadata export, including the keys that connect one table to another.
You can inspect Metabase metadata with AI when the team needs to understand a warehouse without repeatedly looking up the same fields in the interface. The metabase-database-metadata skill retrieves an authorised schema export and arranges it as local files that an agent can search.
Suppose two order tables both have a customer identifier. The table definitions can show which key each field references. That gives an analyst a place to start checking a proposed join, without assuming that matching column labels mean the same thing.
What it does
- Find a database, schema and table by name.
- Read stored descriptions and distinguish field types from their business meanings.
- Trace references to another table or a nested field.
- Keep a local copy for follow-up questions.
- Refresh the cache when the live schema has changed.
Install and use
- Install the skill and the official Metabase metadata package. Live export needs an authorised instance with the serialization endpoint and an API key supplied through your environment.
- Ask “Using the local export, which field connects orders to customers?”
- Keep the cache out of shared version control. Confirm a refresh before replacing the existing derived files.
FAQ
Source
Made by Metabase. Original skill: metabase-database-metadata, published on GitHub under the MIT licence.
View on GitHub