# `PhoenixKit.SchemaPrefix`
[🔗](https://github.com/BeamLabEU/phoenix_kit/blob/v1.7.213/lib/phoenix_kit/schema_prefix.ex#L1)

Applies the configured schema prefix to a PhoenixKit Ecto schema.

`use PhoenixKit.SchemaPrefix` sets `@schema_prefix` from

    config :phoenix_kit, prefix: "myschema"

so every query, insert, preload, join, and bulk operation built on the
schema targets the named Postgres schema the migrations installed into
— without the host having to point the database role's `search_path`
at it. When no prefix is configured (the default `public` install),
`@schema_prefix` compiles to `nil` — equivalent to unset — and
behavior is unchanged.

This is **compile-time** configuration (read via
`Application.compile_env/2` while the host compiles phoenix_kit):
set it in `config/config.exs`, not `runtime.exs`. Mix tracks the value
and recompiles the schemas when it changes. That matches the nature of
the setting — an install's schema prefix is fixed at `mix
phoenix_kit.install --prefix ...` time and never changes at runtime.

A stale build cannot silently target the wrong schema: Elixir records
`compile_env` values and validates them when the application boots
(`Config.Provider.validate_compile_env/1`), so phoenix_kit compiled
before the prefix was configured raises a clear "compiled with a
different value" error instead of quietly querying `public`. Any
`mix compile`/`phx.server`/`release` after the config change
recompiles the schemas automatically.

Every table-backed schema in PhoenixKit must `use` this module right
after `use Ecto.Schema` (a conformance test enforces it). Embedded
schemas don't need it.

---

*Consult [api-reference.md](api-reference.md) for complete listing*