feat(cli): prisma contract print writes the configured contract as Prisma 8 PSL that reads back as the same contract

.agents/rules/README.md

@@ -79,7 +79,7 @@ Rules below are listed by bare filename; the canonical file is `.agents/rules/<n
 
 ## Imports & Layering
 - `import-validation.mdc` — Layering rules and exceptions
-- `no-family-vocabulary-in-framework.mdc` — No family/target vocabulary in packages/1-framework; enforced by the `no-family-vocabulary` Biome plugin + the `lint:framework-vocabulary` ratchet
+- `no-family-vocabulary-in-framework.mdc` — No family/target vocabulary in packages/1-framework, and no name of a Prisma version before 8 outside its CLI; enforced by the `no-family-vocabulary` Biome plugin + the `lint:framework-vocabulary` ratchet
 - `no-inline-imports.mdc` — Prohibit inline type imports in source files
 - `shared-plane-packages.mdc` — Pattern for creating shared plane packages
 - `multi-plane-packages.mdc` — Packages that span multiple planes (shared, migration, runtime)

.agents/rules/no-family-vocabulary-in-framework.mdc

@@ -1,5 +1,5 @@
 ---
-description: The framework domain (packages/1-framework) carries no family- or target-specific vocabulary — types, fields, hooks, or strategy values. Enforced by the no-family-vocabulary Biome plugin plus lint:framework-vocabulary.
+description: The framework domain (packages/1-framework) carries no family- or target-specific vocabulary — types, fields, hooks, or strategy values — and, outside the CLI, no name of a Prisma version before 8. Enforced by the no-family-vocabulary Biome plugin plus lint:framework-vocabulary.
 globs: ["packages/1-framework/**"]
 alwaysApply: false
 ---
@@ -10,12 +10,16 @@ alwaysApply: false
 
 **The test for every new framework export, field, hook, or member:** *would Mongo populate or consume this meaningfully?* If not, it belongs in the family layer (e.g. `@internal/sql-contract`), behind a structural interface + type predicate at the boundary.
 
+The framework also must not know which Prisma version wrote a document: code for an earlier version's schema belongs in the package that reads it (e.g. `@internal/sql-contract-prisma7`). The CLI (`packages/1-framework/3-tooling/cli`) may name such a version, because it talks to users about their projects.
+
 ## How it is enforced
 
-The [`no-family-vocabulary` Biome plugin](../../biome-plugins/no-family-vocabulary.grit) matches syntax nodes — identifiers, string literals, template chunks, import paths — carrying family/target terms (`table`, `column`, `postgres`, `nativeType`, `primaryKey`, …). It emits `info` diagnostics; `pnpm lint:framework-vocabulary` is the ratchet that counts them, deduplicated by file and line, against the `threshold` in `scripts/lint-framework-vocabulary.config.json`. The count may only shrink, and the threshold must be lowered whenever it does.
+The [`no-family-vocabulary` Biome plugin](../../biome-plugins/no-family-vocabulary.grit) matches syntax nodes — identifiers, string literals, template chunks, import paths — carrying family/target terms (`table`, `column`, `postgres`, `nativeType`, `primaryKey`, …) or, outside the CLI, the name of a Prisma version before 8 (`prisma7`, `Prisma 6`, `prisma@5.22.0`, …). `Prisma 8` is never flagged. It emits `info` diagnostics; `pnpm lint:framework-vocabulary` is the ratchet that counts them, deduplicated by file and line, against the `threshold` in `scripts/lint-framework-vocabulary.config.json`. The count may only shrink, and the threshold must be lowered whenever it does.
 
 **Comments and JSDoc are deliberately not counted.** Prose that documents family-blind code correctly ("family-parameterized (SQL, Mongo, …)") is good documentation, and the previous line-scanning checker taxed it — the majority of what it flagged was prose, not surface. A leak that matters is a real type, field, identifier, string, or import path, and the plugin only sees those.
 
+Test files and `test/` directories are not scanned. The CLI's allowance for version names is the only path exemption; do not add another. There are two other exemptions, and no others: a framework-neutral compound listed in the plugin's `not` guards (`SymbolTable`, `MigrationLogTable`, the formatter's alignment columns), and a line suppression.
+
 Suppress a genuinely family-blind site at the line, with a reason:
 
 ```ts

biome-plugins/no-family-vocabulary.grit

@@ -1,5 +1,11 @@
 // Recognise SQL/Mongo/target vocabulary (`table`, `column`, `postgres`,
 // `nativeType`, …) reaching the family-blind `packages/1-framework` domain.
+// Outside the CLI package it also flags the name of any Prisma version before
+// 8 (`prisma7`, `Prisma 6`, `prisma@5.22.0`, …), because the framework reads
+// every PSL document the same way and must not know which product version
+// wrote it. The CLI talks to users about their projects, including a Prisma 7
+// project `orm init` sets Prisma 8 up beside, so it may name the version.
+// `Prisma 8` is never flagged.
 //
 // The scope is enforced by the `$filename` guard below, not by a biome
 // `overrides` entry. Biome 2.5.6 does accept `plugins` inside `overrides` (and
@@ -44,8 +50,8 @@
 // identifier splits on non-alphanumerics and on a lowercase/digit → uppercase
 // hump, then each token is compared case-insensitively, allowing a trailing
 // plural `s`. Reproducing that means the term's own casing decides which
-// boundaries are legal, which is why there are three alternatives instead of
-// one case-insensitive pattern:
+// boundaries are legal, which is why there are three term alternatives instead
+// of one case-insensitive pattern:
 //
 //   1. lowercase head, lowercase tail (`table`) — a lowercase head cannot open
 //      a hump, so the prefix must be empty or end in a non-alphanumeric. This
@@ -59,7 +65,7 @@
 //      stops `SQL` matching inside `SQLQueryPlan`, which the tokenizer reads as
 //      the single token `sqlquery`.
 //
-// Each alternative lists every forbidden term with a literal head character, a
+// Each term alternative lists every forbidden term with a literal head character, a
 // case-insensitive middle, and a literal tail character. The three multi-word
 // terms (`nativeType`, `primaryKey`, `foreignKey`) additionally accept either a
 // separator run or a hump between their two words, so `native-type`,
@@ -101,6 +107,10 @@ or {
   not $filename <: r".*\.(?:test|test-d)\.ts",
   not $filename <: r".*/test/.*\.ts",
   or {
+    and {
+      not $filename <: r".*packages/1-framework/3-tooling/cli/.*",
+      $tok <: r"(?s)(?:(?:|.*[^A-Za-z0-9])p|(?:|.*[^A-Za-z0-9]|.*[a-z0-9])P)(?i:risma)[^A-Za-z0-9]*[1-7](?:|[^A-Za-z0-9].*|[A-Z].*)"
+    },
     $tok <: r"(?s)(?:|.*[^A-Za-z0-9])(?:p(?i:ostgre)s|p(?i:ostgresq)l|s(?i:qlit)e|m(?i:ysq)l|m(?i:ariad)b|m(?i:ssq)l|s(?i:qlserve)r|c(?i:ockroac)h|p(?i:glit)e|m(?i:ong)o|m(?i:ongod)b|c(?i:ollectio)n|c(?i:olum)n|t(?i:abl)e|c(?i:onstrain)t|s(?i:q)l|v(?i:archa)r|j(?i:son)b|d(?i:ialec)t|r(?i:l)s|n(?i:ativ)(?:[eE][^A-Za-z0-9]+[tT]|eT)(?i:yp)e|p(?i:rimar)(?:[yY][^A-Za-z0-9]+[kK]|yK)(?i:e)y|f(?i:oreig)(?:[nN][^A-Za-z0-9]+[kK]|nK)(?i:e)y)[sS]?(?:|[^A-Za-z0-9].*|[A-Z].*)",
     $tok <: r"(?s)(?:|.*[^A-Za-z0-9]|.*[a-z0-9])(?:P(?i:ostgre)s|P(?i:ostgresq)l|S(?i:qlit)e|M(?i:ysq)l|M(?i:ariad)b|M(?i:ssq)l|S(?i:qlserve)r|C(?i:ockroac)h|P(?i:glit)e|M(?i:ong)o|M(?i:ongod)b|C(?i:ollectio)n|C(?i:olum)n|T(?i:abl)e|C(?i:onstrain)t|S(?i:q)l|V(?i:archa)r|J(?i:son)b|D(?i:ialec)t|R(?i:l)s|N(?i:ativ)(?:[eE][^A-Za-z0-9]+[tT]|eT)(?i:yp)e|P(?i:rimar)(?:[yY][^A-Za-z0-9]+[kK]|yK)(?i:e)y|F(?i:oreig)(?:[nN][^A-Za-z0-9]+[kK]|nK)(?i:e)y)[sS]?(?:|[^A-Za-z0-9].*|[A-Z].*)",
     $tok <: r"(?s)(?:|.*[^A-Za-z0-9]|.*[a-z0-9])(?:P(?i:ostgre)S|P(?i:ostgresq)L|S(?i:qlit)E|M(?i:ysq)L|M(?i:ariad)B|M(?i:ssq)L|S(?i:qlserve)R|C(?i:ockroac)H|P(?i:glit)E|M(?i:ong)O|M(?i:ongod)B|C(?i:ollectio)N|C(?i:olum)N|T(?i:abl)E|C(?i:onstrain)T|S(?i:q)L|V(?i:archa)R|J(?i:son)B|D(?i:ialec)T|R(?i:l)S|N(?i:ativ)(?:[eE][^A-Za-z0-9]+[tT]|eT)(?i:yp)E|P(?i:rimar)(?:[yY][^A-Za-z0-9]+[kK]|yK)(?i:e)Y|F(?i:oreig)(?:[nN][^A-Za-z0-9]+[kK]|nK)(?i:e)Y)[sS]?(?:|[^A-Za-z0-9].*)"
@@ -110,5 +120,5 @@ or {
   not $tok <: r"(?s).*(?i:alignment)[^A-Za-z0-9]*(?i:columns).*",
   not $tok <: r"(?s).*(?i:type)[^A-Za-z0-9]*(?i:column).*",
   not $tok <: r"(?s).*(?i:attribute)[^A-Za-z0-9]*(?i:column).*",
-  register_diagnostic(span=$tok, message="no-family-vocabulary: family/target vocabulary in the family-blind framework domain; move the concept into the owning family layer, or suppress with `// biome-ignore lint/plugin/no-family-vocabulary: <why this is family-blind>`", severity="info")
+  register_diagnostic(span=$tok, message="no-family-vocabulary: family, target or product vocabulary in the family-blind framework domain; move the concept into the layer that owns it, or suppress with `// biome-ignore lint/plugin/no-family-vocabulary: <why this is family-blind>`", severity="info")
 }

docs/architecture docs/adrs/ADR 252 - An earlier Prisma version's schema is a contract source.md

@@ -86,15 +86,15 @@ Every lowering rule is checked against SQL that the earlier version's own toolch
 
 ```mermaid
 flowchart LR
-  schema["schema.prisma<br/>(Prisma 7)"] --> parser["@internal/psl-parser<br/>grammar: 'prisma7'"]
+  schema["schema.prisma<br/>(Prisma 7)"] --> parser["@internal/psl-parser<br/>one grammar for every PSL document"]
   parser --> interpreter["@internal/sql-contract-prisma7<br/>rules of the Prisma 7 language for the SQL family"]
   binding["@internal/target-postgres<br/>prisma7PostgresBinding: what Postgres creates"] --> interpreter
   interpreter --> contract["Contract"]
   facade["@prisma/orm-postgres/config<br/>prisma7Schema(path)"] -. wires .-> interpreter
   facade -. wires .-> binding
 ```
 
-**The parser is shared, and the earlier grammar is opt-in.** `@internal/psl-parser` reads both languages. The two additions the earlier language needs, attributes on enum members and field lines inside a `view` block, are read only under the `grammar: 'prisma7'` parse option. The default grammar is unchanged, so a Prisma 8 schema keeps rejecting exactly what it rejected before, and nothing here reaches a user who never adopts the reader.
+**The parser is shared, and it has one grammar.** `@internal/psl-parser` parses every PSL document the same way, whichever Prisma version wrote it. The two constructs the earlier language needs, attributes on enum members and field lines inside a `view` block, parse in every document. Each reader decides what it accepts. The Prisma 8 readers report an attribute on an enum member, and report a `view` block like any other block whose keyword no composed pack claims, so a Prisma 8 PSL contract that uses either fails. The Prisma 7 reader reads enum member attributes and refuses a `view` with the error shown above.
 
 **Rules of the language live in the family authoring package.** `@internal/sql-contract-prisma7` holds everything that is true of the Prisma 7 language for the SQL family: blocks and attributes, relation pairing, junction tables, defaults, and the diagnostics. It knows nothing about a particular database and depends on no Prisma 7 package.