capy-auth 0.10.0

Optional database-backed browser sessions and explicit authorization for Capy.
Documentation
# capy-auth

Optional conventional browser authentication for Capy 0.10.0. Applications depend on
`capy-framework` with the `auth` feature and install it using `capy make:auth`.
The public auth surface remains experimental.

```sh
cargo install capy-cli --version 0.10.0 --locked
capy new notebook
cd notebook
capy make:auth
capy migrate
capy types
capy dev
```

Server-side sessions, Argon2id passwords, synchronizer CSRF and explicit policies.
Native SeaORM and Axum remain available. SQLite and PostgreSQL use the same
conventional migration; API-only projects require no Node or React.

Session/reset credentials have 256 bits of OS randomness and are stored as SHA-256
digests. Production cookies are HttpOnly, Secure, SameSite=Lax and host-only; local
HTTP has deliberate development defaults. Login rotates the session; logout
revokes it. Password reset consumes an expiring single-use token and revokes all
sessions. Required/optional identity extractors reject malformed credentials.

Production requires HTTPS and an application-owned `ResetDelivery` implementation.
The generated private local inbox and memory adapter are development-only. Auth
protects unsafe routes with a synchronizer token; the generated React adapter
integrates with the existing Fetch helper. Throttling is process-local and uses
the socket peer; deployments behind proxies or across replicas need edge controls.
This crate is an ordinary web-app identity option, not an external identity provider.

See the [API documentation](https://docs.rs/capy-auth/0.10.0/capy_auth/),
[authentication guide](https://github.com/capy-rs/capy/blob/v0.10.0/docs/authentication.md)
and [notebook tutorial](https://github.com/capy-rs/capy/blob/v0.10.0/docs/auth-tutorial.md).
Repository links require authorized access while the repository is private.