Catalog

The Catalog enables organizations to preserve and share reusable queries and datasets through versioned Entities. Entities combine code, metadata, linked data sources, and execution records.

Entity Components

Each entity contains:

• Metadata — Title, description, tags, type (model or metric)
• Code — SQL or Python code
• Structure — Column definitions and data types
• Data Source — Connected database or warehouse
• Execution History — Last run timestamps and results

Saving Entities

• Admins can save directly as Draft or Published
• Team members can submit suggestions for admin review (private until approved)

Editing and Execution

• Preview mode — Execute code without saving changes
• Run and Save — Execute code and record results with updated timestamps

Lifecycle

For Members

1. Save entities from reports
2. Await admin approval
3. Optionally withdraw suggestions

For Admins

• Bypass approval workflow
• Publish or modify entities directly
• Review and approve member submissions

Access Control

Users only see entities whose data sources they can access. Published and approved entities are visible to authorized users, while drafts and archived items are restricted to owners and admins.

AI Integration

MetricChat indexes entity attributes — titles, descriptions, tags, code structure, and column information — to enable semantic discovery. The AI can recommend relevant catalog entities during analysis.