GraphQL Nullability Contract Repair

The problem

GraphQL schemas often promise non-null values that the underlying database, resolvers, or loaders cannot actually guarantee. This drift leads to runtime errors, silent data bubbling that nulls out entire parent objects, and brittle client-side assertions that crash when data is missing or deleted.

What it does

• Traces field nullability from database columns through ORMs, loaders, and resolvers to the final SDL.
• Identifies list-container and item-nullability mismatches that cause entire arrays to fail when one item is missing.
• Detects "catch-and-null" resolver patterns that hide operational failures behind legitimate absence.
• Repairs loaders to ensure they preserve key order and cardinality instead of silently dropping missing records.
• Generates safe repository patches for input presence, subscription payloads, and normalized cache policies.

Why this beats prompting it yourself

Manual review usually misses the propagation path where a single non-null violation bubbles up the tree. This skill systematically audits the entire chain from persistence to generated client types, ensuring that a fix in the schema is backed by real resolver logic and test fixtures.

Use cases

• Fixing non-null fields that are actually nullable in legacy database rows.
• Resolving loader bugs where missing keys cause array length mismatches.
• Differentiating between omitted input fields and explicit nulls in mutations.
• Updating stale generated types and removing unsafe non-null assertions in UI code.

Known limitations

Does not access live production APIs or databases. It cannot make product-level decisions on whether a field should be nullable, but it identifies where a decision is required.