Map database columns with AI
Document which source fields feed a destination table, where values change and which fields still need a decision before migration.
Map database columns with AI when two systems disagree about what a field means. With schema-mapper, an analyst can turn two schema inventories into a mapping that a colleague can review. A price in whole pennies and a price in pounds might share a name and a numeric type. The conversion still belongs in the document.
The accompanying comparison utility finds matching names and differences in declared types. For a rename, a calculation or a value assembled from several fields, the agent needs a definition or an existing transformation rule. The finished mapping records those decisions alongside any gaps.
What it does
- Identify destination fields that have no corresponding source.
- Record the arithmetic needed to convert units or currencies at a supplied rate.
- Mark fields the migration intentionally drops.
- Spot a change from a timestamp to a date that could discard information.
- Give reviewers a list of unresolved mappings and the person or team who can settle each one.
Install and use
- Install the skill in a compatible agent. Use Python 3 and two local schema CSV files for the comparison tool. Include the column names and data types.
- Ask, “Compare the old and new order schemas. Flag every monetary field where the destination unit is unknown.”
- Review the suggested matches, add the transformation rules and agree on defaults before implementing the migration.
FAQ
Source
Made by Nimrod Fisher. Original skill: schema-mapper, published on GitHub under the MIT licence.
View on GitHub