Metadata

Adding Metadata to data sources

Wobby uses column-level metadata to help AI agents understand your data better. Providing clear descriptions and constraints improves the accuracy of responses when users ask questions about your data.

You can add metadata manually or upload it using a YAML file (such as a dbt schema.yml).

---

Column Metadata Fields

Each column in a table supports the following metadata fields:

• Description Briefly explain what this column represents in the business context.

• Classification (Optional) Label the column with types like identifier, PII, or financial to help the AI interpret it correctly.

• Data Type This is auto-detected but visible to you (e.g. INTEGER, DOUBLE, VARCHAR, etc.).

• Constraints Define any rules the column must follow (e.g. Primary key, Not null, Date must be in the past, Status must be OPEN/CLOSED, etc.).

You can enter this metadata manually by selecting a column and filling in the fields under the Metadata tab.

---

Import Metadata from YAML

To speed things up, you can upload a YAML file to populate metadata for a table.

How to Upload

1. Go to the Metadata tab for a table.

2. Click Import Metadata.

3. Drag and drop your .yml or .yaml file, or paste the YAML content directly into the input box.

4. Click Upload to apply the metadata.

YAML Format Example

Make sure your YAML file follows this structure:

version: 2
models:
  - name: lh_suppliers
    description: Description of the table
    columns:
      - name: column_name
        description: Description of the column
        constraints: Primary key, Not null
        classification: identifier

✅ Tip: If you're using dbt, you can reuse your existing schema.yml.

---

Best Practices

• Add descriptions for every column—even simple ones like created_at or is_active. The more context the AI has, the better.

• Include constraints wherever they are known to improve the agent’s understanding of data rules.

• Test the Agent's outputs and refine metadata based on the Agent's reasoning.