This is the Prisma Next feature-support matrix feeding PR #986 (the Prisma 8 RC1 release). Each row is one feature; the three data columns state whether that feature is available in Prisma Next for Postgres, SQLite, and MongoDB — not whether Prisma 7 had it. The matrix lists all features Prisma Next exposes (mined from prisma-8-api.md), plus every notable Prisma 7 feature Prisma Next lacks, so that absences are named rather than silently missing. The Prisma Next evidence column points only to Prisma Next (TypeScript) test suites; Rust/engine tests and Prisma 7 tests are never cited here. When a cell is 🟡 the evidence is intentionally empty — that emptiness is what makes it untested rather than proven.
The matrix is split by category across the scorecard/ directory — this page is the index. Each category file is self-contained (it repeats the legend below) and holds that category's table(s) verbatim. Follow the links in the Categories section to reach a specific category.
Legend:
✅Works — proven by a Prisma Next integration test (one that executes the feature against a database — Postgres via PGlite, SQLite via its real driver, or MongoDB via mongodb-memory-server — and asserts the observable runtime result). Unit-tier tests (SQL/AST/plan/type/snapshot assertions, or any test that never hits a database) do not qualify. Per-database rigor applies: a Postgres integration test cannot justify a SQLite or MongoDB✅, and vice versa.🟡Untested — reachable through the Prisma Next public surface, but no proving Prisma Next integration test exists yet (evidence left blank). This includes features whose only backing is a unit-tier test.🧪Experimental — shipped in Prisma Next but outside the stability promise (polymorphism / multi-table inheritance).❌Not in 8.0 — deliberately absent from Prisma Next.—n/a — feature does not apply to that database.
Each row names exactly one capability. Where an operator/command/stage set was previously bundled, it has been split so every distinct feature carries its own per-database verdict and its own evidence.
| Category | Feature rows |
|---|---|
| Targets & connection | 14 |
| Types & values | 15 |
| PSL schema language | 44 |
| Contract emission & authoring | 14 |
| SQL query builder | 46 |
| SQL ORM client | 52 |
| MongoDB query & ORM | 81 |
| Relations | 20 |
| Filtering | 24 |
| Ordering & pagination | 11 |
| Aggregation & grouping | 20 |
| Nested writes & atomic ops | 25 |
| Raw & typed SQL | 9 |
| Transactions | 9 |
| Migrations | 123 |
Introspection (contract infer) |
26 |
| CLI commands | 32 |
| Extensions | 15 |
| Observability & lifecycle | 13 |
The Migrations file gathers all twelve Migrations — * sub-topics (workflow; columns & types; IDs, PKs & autoincrement; foreign keys; indexes & unique; enums; defaults; native types; extensions; views; existing-data safety; schema filters) as sub-headings within a single file.
Group structure (30 groups): Targets & connection; Types & values; PSL schema language; Contract emission & authoring; SQL query builder; SQL ORM client; MongoDB query & ORM; Relations; Filtering; Ordering & pagination; Aggregation & grouping; Nested writes & atomic ops; Raw & typed SQL; Transactions; Migrations — workflow; Migrations — columns & types; Migrations — IDs, PKs & autoincrement; Migrations — foreign keys; Migrations — indexes & unique; Migrations — enums; Migrations — defaults; Migrations — native types; Migrations — extensions; Migrations — views; Migrations — existing-data safety; Migrations — schema filters; Introspection (contract infer); CLI commands; Extensions; Observability & lifecycle.
Verdicts are computed per feature row across the three DB columns; the tallies below count individual non-— cells.
Across 593 atomic feature rows (one capability each), spanning 1,779 per-database cells (593 rows × 3 databases): ✅ 416, 🟡 488, 🧪 12, ❌ 244 (and 619 — n/a cells).
Per-database tallies:
| Database | ✅ | 🟡 | 🧪 | ❌ | — |
|---|---|---|---|---|---|
| Postgres | 222 | 136 | 4 | 111 | 120 |
| SQLite | 85 | 235 | 4 | 98 | 171 |
| MongoDB | 109 | 117 | 4 | 35 | 328 |