Resuma — resumable SSR web framework for Rust

Instantly-interactive Rust web apps — without hydration. Not a résumé/CV builder.

Ship HTML from the server. The browser resumes serialized state and lazy handler chunks — it never re-runs your components. Static pages ship zero client JS.

Official repo: github.com/GoldevLab/resuma · Docs: resuma-docs.fly.dev · Crate: crates.io/crates/resuma

Get started · Benchmark · Docs · API

cargo install resuma
resuma new my-app --template todo
cd my-app && resuma dev --open

Measured: smallest interactive runtime in class

Same UX everywhere — SSR counter (heading + increment button). Production builds, gzip transfer sizes. Full methodology →

Regenerate locally: node benchmark/run.mjs (or --resuma-only to refresh just Resuma) · Independent sources in benchmark/README.md

If you like Qwik's resumability and a Rust-first stack, Resuma is that model with native SSR, server actions, and no WASM bundle by default.

Why Resuma?

Mental model: components run once on the server. SSR embeds signals and handler refs in HTML; a tiny client runtime resumes on first click or when an island enters the viewport. Multiplayer apps can use resuma::realtime for WebSocket rooms and peers.

Quick start

Requires: Rust 1.91+ (rustup)

cargo install resuma
resuma new                          # interactive — name + template menu
resuma new my-app --template todo   # full showcase (signals, server, islands)
cd my-app
resuma dev --open

Templates: basic · todo · flow · flow-booking · flow-fullstack · production

AI assistants (Cursor, Codex, Gemini)

Teach your editor Resuma patterns — view!, Flow, HtmlTheme / Popup / Modal, SeoKit, deploy — with one command:

resuma install skill              # ~/.cursor/skills/resuma/ (global)
resuma install skill --project    # .cursor/skills/resuma/ (this repo)
resuma install skill --target agents  # ~/.agents/skills/ (Codex-style)
resuma install skill --force      # overwrite after a CLI upgrade

Skill vs MCP: start with the skill (guidance for writing code). An official MCP for docs search / CLI tools may come later.
Guide: resuma-docs.fly.dev/docs/integrations/ai_assistant

use resuma::prelude::*;

#[component]
fn Counter() {
    let count = signal(0);
    view! {
        <main>
            <h1>"Count: " {count}</h1>
            <button onClick={count.update(|c| *c += 1)}>"+"</button>
        </main>
    }
}

Handlers compile to JavaScript automatically — lazy-loaded on first interaction, wired to resumed signal state.

Library only:

[dependencies]
resuma = "1.3"
tokio  = { version = "1", features = ["full"] }

Server actions

#[data]
struct SearchResult {
    title: String,
    url: String,
}

#[server]
async fn search(q: String) -> Vec<SearchResult> {
    db::search(&q).await
}

#[server] registers POST /_resuma/action/search. Call from the client with js! or Flow forms — no custom wiring.

Resuma Flow

Full-stack layer inside the same crate — file-based pages, loaders, submits (like Qwik City or SolidStart):

resuma new my-app --template flow
resuma add sqlx    # optional DB scaffolding

Guide: resuma-docs.fly.dev/docs/flow

Product map

Details: docs/NAMING.md

CLI

Architecture

┌──────────────────────────────────────────────────────────┐
│                   resuma crate                           │
│   core → ssr → server (axum) → flow + router             │
│   resuma-macros (view!, #[component], rs2js → JS)       │
└───────────────────────────┬──────────────────────────────┘
                            │ HTTP
                            ▼
┌──────────────────────────────────────────────────────────┐
│              Browser (~1 KiB loader, lazy core)          │
│   parse resuma/state · delegate events · lazy handlers   │
└──────────────────────────────────────────────────────────┘

Deep dive: docs/ARCHITECTURE.md · Security: docs/SECURITY.md · Backend patterns: docs/BACKEND.md

Resources

Project layout

resuma/
├── assets/               # logo + GitHub social preview
├── crates/
│   ├── resuma/           # runtime, SSR, server, flow, CLI
│   ├── resuma-macros/    # view!, #[component], rs2js
│   └── resuma-flow/      # Flow UI widgets (flow_graph, dashboard, worker_panel)
├── runtime/              # TypeScript client (~1 KiB loader; see benchmark/)
├── benchmark/            # measured comparisons vs Qwik, Next, etc.
├── examples/             # runnable demos
└── docs/                 # markdown reference

Docs site (separate repo): resuma-docs.fly.dev

From source:

git clone https://github.com/GoldevLab/resuma
cd resuma
cargo install --path crates/resuma --features cli --force
cargo run -p example-counter
cargo run -p example-todo

Manual apps can mount no-props components without the verbose render path:

ResumaApp::new()
    .component("/", App)
    .serve(ServeOptions::default())
    .await

What ships in 1.3

• view! / #[component] — JSX-like templates; resumable by default
• #[server] / #[submit] / #[load] — server RPC, form mutations, and Flow data loaders
• #[island] — optional lazy client bundles
• SPA NavLink + view transitions
• ResumaError::Redirect + middleware auth gating
• Cookies / session helpers (Redirect::with_session_cookie)
• Workers, artifacts, and uploads
• resuma::realtime — WebSocket rooms / peers
• WebGPU CSP defaults
• Client asset cache-busting
• Per-action rate limits
• Flow file-based routing (FlowApp, src/pages/)
• CLI: new, add, dev, build, routes, update, doctor

Community

• Issues & ideas: GitHub Issues
• Security: see SECURITY.md — please do not file public issues for exploitable bugs

Contributions welcome — read CONTRIBUTING.md first.

Why "Resuma"?

Spanish for both resumes (continues) and summary — the framework resumes execution from a serialized summary of the server-side render.

Looking for a resume/CV app? This repository is a Rust full-stack web framework (SSR, Axum, server actions). Resume-builder projects on GitHub use the same name but are unrelated.

Find this project (search & links)

Topics: Rust · SSR · resumability · Axum · server actions · islands · zero hydration · resuma on crates.io.

License

Dual-licensed under MIT OR Apache-2.0.