feat: move node-type-registry to packages/ and add column-ref annotations

[pyramation]

Summary

Two changes in one PR:

1. Move graphql/node-type-registry → packages/node-type-registry

The node-type-registry is platform-level infrastructure (used by blueprints, codegen, and the SQL metaschema), not graphql-specific. Moves it alongside packages/safegres where it belongs. The pnpm-workspace.yaml already includes packages/*, so no workspace config changes needed.

2. Add "format": "column-ref" JSON Schema annotations

Every parameter_schema property that references a table column name now carries "format": "column-ref". This is a standard JSON Schema annotation that distinguishes column references from config values (booleans, enums, integers, table/schema names, UUIDs).

Annotated across all node type categories:

• Authz* (16 types): entity_field, is_published_field, published_at_field, valid_from_field, valid_until_field, array_field, anchor_field, owner_field, obj_field, obj_ref_field, owned_table_key, owned_table_ref_key, this_object_key
• Data* (19 types): field_name, owner_field_name, entity_field_name, target, source_fields, parent_fk_field, fields, source_field_name, role_key_field_name, protected_field_names, condition_field, watch_fields, payload_fields
• Search* (7 types): field_name, source_fields, source_geom_field, source_fk_field, trgm_fields, boost_recency_field
• Relation* (5 types): field_name, source_field_name, target_field_name
• View* (5 types): group_by_fields, field_names, primary_columns, primary_field, join_field, columns

Why: The downstream constructive-db SQL codegen package (PR #967) compiles these definitions into a metaschema_public.node_type_registry SQL table. With format: "column-ref" annotations, any SQL trigger (like embedding_chunks) can generically discover which policy parameters are column references and clone the corresponding columns — without hardcoding knowledge of specific Authz*/Data* types. This is the key enabler for issue #767.

Review & Testing Checklist for Human

• Verify the move doesn't break any imports in downstream packages that depend on node-type-registry by npm name (should be transparent since package.json name is unchanged)
• Spot-check a few format: "column-ref" annotations to confirm the right properties were annotated (column refs) and the right ones were left alone (config values like booleans, enums, UUIDs)
• Run pnpm build in packages/node-type-registry to verify TypeScript compilation

Notes

• The format: "column-ref" annotation follows JSON Schema conventions — format is the standard keyword for semantic type hints
• For array-of-column-refs (e.g., entity_fields, source_fields), the annotation is placed on the items schema, not the array itself
• Types with no column-ref properties (AuthzAllowAll, AuthzDenyAll, AuthzComposite, DataTimestamps, DataSoftDelete, etc.) are left unchanged
• After this merges, the constructive-db node-type-registry codegen package needs to be regenerated to pick up the new annotations

Link to Devin session: https://app.devin.ai/sessions/20dbaaa0e6e74599842fbefe33efbc26
Requested by: @pyramation

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring

[socket-security]

Review the following changes in direct dependencies. Learn more about Socket for GitHub.

Diff	Package	Supply Chain Security	Vulnerability	Quality	Maintenance	License
	pgpm@​1.4.2

View full report