runique 3.0.2

A Django-inspired web framework for Rust with ORM, templates, and comprehensive security middleware
Documentation

Runique β€” the Django developer experience, in type-safe Rust

Rust Tests passing License Version Crates.io GitHub stars Runique

Declare a model once, and you get the database table, the migration, a type-safe form, and a full admin panel β€” no extra wiring. Runique brings Django's productivity to Rust without asking you to give up Rust's safety or performance. It's built on Axum, SeaORM and Tera, and it stays out of your way once the boilerplate is gone.

Status, plainly: active development. The framework crate (runique) is the source of truth; demo-app is a real application exercised against it, not a toy. The admin panel is in beta. Nothing below is dressed up β€” see the project status for the unfiltered version.

🌍 Languages: English | Français


Declarative macros, not boilerplate

model! {
    Article,
    table: "articles",
    pk: id => Pk,
    enums: { Status: [Draft="Draft", Published="Published"], },
    {
        title:  text [required],
        slug:   text [unique],
        body:   richtext [required],
        status: choice [enum(Status), default: "Draft"],
        views:  int [default: 0],
    }
}

model! generates the SeaORM entity (article::Model) and its SQL migration (runique makemigrations) from the same declaration. Pair it with #[form] and you get a matching type-safe form, validated server-side and derivable straight from the schema. Register the resource in admin! and the CRUD panel is already there β€” list view, search, filters, permissions, all of it:

admin! {
    article: article::Model => ArticleForm {
        title: "Articles",
        list_display: [["title", "Title"], ["status", "Status"], ["views", "Views"]],
        list_filter:   [["status", "Status", 5]],
    }
}

Why Runique

Rust already has fast, low-level building blocks for the web β€” what it doesn't have is a framework that gives you Django's day-to-day productivity out of the box. Wiring an ORM, a template engine, a forms layer and an admin together yourself is a project of its own before you've written a single feature. Runique does that wiring for you, following one set of conventions, so the time goes into your app instead of your plumbing β€” and you keep Rust's type safety and performance the whole way through.

Django (Python) Runique (Rust)
models.py model! β†’ SeaORM entity + migration
forms.py #[form] type-safe forms
admin.py admin! generated admin panel
urls.py urlpatterns! routing macro
Django templates Tera (auto-escaped)
QuerySet SeaORM + search! query DSL
middleware ordered middleware slots

For the full picture: Runique vs Django.


Security by default

None of this is bolted on afterward β€” it's part of the base you start from:

  • CSRF protection compares tokens in constant time (ct_eq), so timing can't leak a match
  • CSP ships with a per-response nonce, configurable through the builder
  • Login is timing-safe (no user enumeration through response time) and passwords are hashed with Argon2
  • Sessions persist to the database, with authenticated sessions protected first when memory runs low
  • Password reset tokens live in the database hashed with SHA-256, single-use, and hardened against IDOR
  • Output is sanitized (ammonia) on top of Tera's own auto-escaping, and host validation is enforced

Security policy


Quick start

runique new myapp
cd myapp
cargo run            # your app is a normal Rust binary

runique start regenerates the admin CRUD code from your admin! declarations, then launches cargo run itself β€” it's a one-shot generation step chained into the launch, not a background watcher (see Admin (beta)). Plain cargo run skips regeneration.

A trimmed-down main.rs (the full version lives in demo-app/src/main.rs):

use runique::prelude::*;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let config = RuniqueConfig::from_env();
    let db = DatabaseConfig::from_env()?.build().connect().await?;

    RuniqueApp::builder(config)
        .routes(url::routes())
        .with_database(db)
        .statics()
        .build()
        .await
        .map_err(|e| -> Box<dyn std::error::Error> { Box::new(e) })?
        .run()
        .await?;
    Ok(())
}

Routes go through the urlpatterns! macro and come out as a regular Axum Router:

pub fn routes() -> Router {
    urlpatterns! {
        "/"          => view!{ index },        name = "index",
        "/blog/{id}" => view!{ blog_detail },  name = "blog_detail",
    }
    .rate_limit("/login", "login", view!(login_user), 10, 60, vec![Method::POST])
}

For the full walkthrough: Installation


What's in this repository

  • runique/ β€” the framework crate itself, the product and the source of truth
  • demo-app/ β€” a real application built against the framework, used to validate it
  • docs/ β€” documentation in English and French

Workspace version (source of truth): 3.0.2.


CLI

runique gives you:

  • runique new <name>
  • runique start [--main src/main.rs] [--admin src/admin.rs] β€” regenerates admin code, then launches the app (one-shot, not a watcher)
  • runique create-superuser
  • runique makemigrations --entities src/entities --migrations migration/src [--force false]
  • runique migration up|down|status --migrations migration/src

⚠️ A note on rolling back migrations runique makemigrations writes migrations that keep the chronological order of the migration system intact. If you ever need to roll one back, reach for the SeaORM CLI instead β€” it keeps the migration tracking table in sync with the schema's actual state. Mixing the two rollback paths can desynchronize that tracking.


Admin (beta)

runique start does three things, in order, on a single thread:

  1. parses your admin! declarations in src/admin.rs
  2. generates the CRUD code under src/admins/
  3. runs cargo run --release, blocking

It checks for .with_admin(...) in src/main.rs first and only generates/launches if that's present β€” otherwise it exits with a message telling you why. There's no continuous watching: run runique start again to regenerate after editing src/admin.rs.

It's still beta: permissions work mainly at the resource level for now, the generated src/admins/ folder gets overwritten on each regeneration, and hardening is ongoing rather than finished.

Admin docs: Admin


Features and database backends

Enabled by default: orm, which brings SeaORM (entities, queries, migrations, sessions) but no database driver: a build with orm alone compiles, and makemigrations works, but the app can't connect to anything.

Pick exactly one backend explicitly: sqlite, postgres, mysql, mariadb (mutually exclusive β€” enabling two at once is a compile error). Each one enables orm too, so features = ["postgres"] is enough. The all-databases feature stays available for multi-engine tooling that needs to talk to every backend at once.


Sessions

CleaningMemoryStore stands in for the default MemoryStore, adding automatic cleanup of expired sessions, a two-tier watermark (128 MB / 256 MB) to keep memory bounded, and priority for authenticated sessions β€” they're the last to be purged under pressure, and they survive restarts through a database fallback.

Full reference: Sessions


Tests and coverage

  • Tests reported: 2875 passing (110 ignored)
  • Coverage snapshot (2026-10-08, package runique, admin module included): functions 88.72%, lines 88.81%, regions 87.25%
cargo llvm-cov --package runique --features all-databases --summary-only

Full per-file breakdown: docs/couverture_test.md


Documentation


Project status & resources


License

MIT β€” see LICENSE