How rusty works

rusty is organized around a native pipeline. Cargo is an interop boundary during migration, not the long-term implementation of resolve, lock, or test.

Preferred pipeline

manifest (rusty.json or Cargo.toml)
  → native resolve
  → lock (rusty.lock or Cargo.lock)
  → fetch (index + checksum)
  → CAS ingest / materialize
  → native rustc graph (build / test)
  → layers + artifact cache

Contrast with a cargo-wrapper mental model: rusty does not aim to shell out to cargo for correctness of core package-manager or build verbs.

Major subsystems

Subsystem	Role
CLI	Thin dispatch for env, PM, store, layers, cache, workspace, auth
Package manager	Manifests, resolve, lock, fetch, add/rm/update, build driver, tests, migrate
Toolchain	Channel install, active toolchain, shims under `$RUSTY_HOME/bin`
Content store	Content-addressable blobs and crate trees
Cache / layers	Artifact groups and per-crate differential layers (`.rusty/`)
Workspace	Monorepo member discovery, affected packages, dependents
Auth	Optional product identity (Clerk), separate from crates.io

Dual-mode projects

Mode	Manifest / lock	Intent
rusty-format	`rusty.json` + `rusty.lock`	Native-first; fail closed when unsupported
cargo-format	`Cargo.toml` + `Cargo.lock`	Interop while migrating; native where implemented

Detection uses those files plus .rusty/format. See Project formats.

Build philosophy

build / run / check / test: native graph or hard error.
Lifecycle scripts: optional overrides in rusty.json (scripts.build, scripts.test, …).
Cargo interop verbs: tiny allowlist only; documented as needing cargo on PATH.
No silent online cargo fallback for rusty-format when native prep fails.

Where data lives

Location	Contents
`$RUSTY_HOME`	Toolchains, global store, auth, bin shims
`<workspace>/.rusty/`	Format marker, target dir, layer index, project caches

Details: Store and cache, Environment.

Get started

Core concepts

Guides

Reference

Preferred pipeline

Major subsystems

Dual-mode projects

Build philosophy

Where data lives

​Preferred pipeline

​Major subsystems

​Dual-mode projects

​Build philosophy

​Where data lives

Preferred pipeline

Major subsystems

Dual-mode projects

Build philosophy

Where data lives