// the find
prisma/prisma-engines
🚂 Engine components of Prisma ORM
This is the Rust internals that power Prisma ORM: the schema engine (migrations, introspection), the PSL schema-language parser, and — as of the v7 architecture — a query compiler that emits execution plans run by TypeScript driver adapters instead of a monolithic query engine binary. It's for people contributing to Prisma itself or debugging why a migration or query behaves a certain way, not for typical app developers consuming Prisma Client.
The split between Rust query compiler and TS driver adapters is a real architectural bet — plans are generated natively and executed via JS drivers, which sidesteps shipping a native binary per platform/runtime combo (Wasm bundle instead). Integration tests actually spin up Postgres, CockroachDB, MSSQL, MySQL/MariaDB and SQLite variants through docker-compose and nextest rather than mocking the database layer. The repo documents its own module boundaries unusually well for an internals repo — query-compiler vs query-core vs query-structure vs schema-engine are each clearly scoped in the README, not left to reverse-engineering.
Onboarding is heavy: direnv/nix, a pinned Rust toolchain, Node 20+ with pnpm, Docker, and a sibling checkout of prisma/orm just to get driver adapters symlinked correctly — this is not a clone-and-cargo-build repo. There's no standalone query engine binary anymore, so understanding a single query execution path means jumping between Rust crates and a separate TypeScript executor process, which raises the bar for first-time contributors. The README itself flags that quaint, schema-engine, and connector-test-kit test setup is under-documented ('help needed' with no guide). Windows support for the driver-adapter test harness is explicitly manual-only — you're re-implementing Makefile steps by hand outside WSL.