fix: only treat @db.Uuid columns as UUIDs — drop field-name inference

[samuelho-dev]

Ports the UUID-detection fix (already merged to release/next as #37, published in 6.0.0-next.4) to the stable main line.

Problem

isUuidField had a third tier that inferred UUID from the field name (/^id$/, /_id$/, /^.*_uuid$/, /^uuid$/) for any String column. UUID is a column type, not a naming convention — the DMMF already knows the type. Every text identifier ending in _id (Stripe acct_/cus_/txn_/…, slugs, provider refs) got a spurious Schema.String.check(Schema.isUUID()) and Died at decode ("Expected a UUID, got acct_…").

Fix

Remove the name tier. isUuidField returns true only for @db.Uuid (native type or documentation) — which captures 100% of genuine UUID columns. Test + README + CLAUDE updated; changeset added (minor).

✅ 347 tests pass, typecheck + build clean.

Summary by CodeRabbit

• Bug Fixes

  • UUID detection is now more accurate, only recognizing fields explicitly marked with @db.Uuid in Prisma. This removes false positives from field name inference that could cause issues with non-UUID text identifiers.

• Documentation

  • Updated documentation on UUID field detection and how to explicitly mark UUID columns using @db.Uuid or custom type annotations.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: c8dc14a0-f610-43d3-885d-a52e5146b8f1

📥 Commits

Reviewing files that changed from the base of the PR and between c965256 and 65bd59e.

📒 Files selected for processing (5)

• .changeset/uuid-detection-no-name-inference.md
• CLAUDE.md
• README.md
• src/__tests__/prisma-parsing.test.ts
• src/prisma/type.ts

Disabled knowledge base sources:

• Linear integration is disabled

You can enable these sources in your CodeRabbit configuration.

---

📝 Walkthrough

Walkthrough

UUID detection in prisma-effect-kysely is simplified to rely exclusively on Prisma type information via @db.Uuid in native type or field documentation. Field-name pattern inference (id, _id, _uuid, uuid) is removed to eliminate false positives from external identifiers like Stripe IDs. Core logic, tests, and documentation are updated together.

Changes

UUID Detection Simplification

Layer / File(s)	Summary
Core UUID detection logic src/prisma/type.ts	isUuidField function now checks only field.nativeType[0] === 'Uuid' and field.documentation containing @db.Uuid; name-pattern fallback logic and String-type gating removed.
UUID detection test coverage src/__tests__/prisma-parsing.test.ts	Test suite reworded to reflect authoritative @db.Uuid detection and replaced field-name-pattern test with explicit assertions that common naming conventions (_id, _uuid, id, uuid) do not trigger UUID inference.
Documentation and changelog CLAUDE.md, README.md, .changeset/uuid-detection-no-name-inference.md	Internal dev guidance, user-facing README, and changeset documentation updated to define UUID detection as Prisma-type-only, remove name-pattern tier, and provide migration guidance via /// @db.Uuid`` or @customType(...) for explicit UUID marking.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/uuid-detection-no-name-inference-main

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}