An encrypted time-of-day field: Encryptor.Ecto.Binary over the value's ISO 8601 form (ADR-0001 decision 1).

defmodule Payments.Encrypted.Time do
  use Encryptor.Ecto.Time, vault: Payments.Vault
end

defmodule Payments.Cards.Card do
  use Ecto.Schema

  schema "cards" do
    field :merchant_id, :string
    field :contact_window_opens_at, Payments.Encrypted.Time
  end
end

Everything below the cast/parse pair is Encryptor.Ecto.Binary, called rather than copied: the same closed option set, the same declared "table"/"column" context, the same tenant resolution, the same :binary column, the same exception family, and the vault's bytes stored verbatim. Read that module for all of it; only the differences are documented here.

The cast/parse pair

cast/2 is Ecto.Type.cast(:time, value) - Ecto's own caster - and the plaintext is Time.to_iso8601/1, parsed back with Time.from_iso8601/1.

The textual form is not a choice taken here. Cloak.Ecto.Time writes the same bytes: cloak_ecto 1.3.0 serializes every scalar with to_string/1, which for a Time is byte-identical to Time.to_iso8601/1. ADR-0004's migration hands the migrator a legacy plaintext and re-encrypts it verbatim below the schema layer (ADR-0002 decision 3), so a column arriving from Cloak.Ecto.Time is readable through this type because the two agree about what the bytes say. The two datetime types are the exception - see Encryptor.Ecto.NaiveDateTime - and every kind's exact bytes are pinned in test/encryptor/ecto/scalar_types_test.exs.

A payload that decrypts and then does not parse raises Encryptor.Ecto.SerializationError with {:unparsable, :time}: the decrypt succeeded, so it is neither an encryption failure nor an integrity event, which is the row ADR-0001 decision 6 gives that exception. A value that reaches dump/3 without passing cast/2 - through insert_all/3, say - raises ArgumentError naming the shape it was given and never the value.

Sub-second precision is dropped at the cast

:time is the primitive this type wraps, and Ecto's caster truncates a microsecond field to {0, 0} on the way in, exactly as it does for a plain :time column. The truncation therefore happens before any encryption and is visible in the changeset rather than only on the next read. A host that needs sub-second precision is not asking for :time and should say so with its own encoding over Encryptor.Ecto.String.

Everything else

nil is NULL and is not encrypted (decision 7). The column is :binary, not :time - type/1 returns :binary whatever the plaintext was (decision 2). Encrypted columns are not queryable, sortable or uniquely indexable (decision 10).