What problem does it solve? Rust code written for the GRID Craft Launcher often drifts from the project's established patterns, producing inconsistent error types, ad-hoc HTTP clients, blocking calls on the async runtime, and unvetted dependencies. This Skill gives an AI assistant the project's exact rules before it writes any Rust in gcl-core or gcl-cli. ## Core Features & Use Cases - Workspace layout rules: gcl-core as a library with one directory per module, gcl-cli mapping subcommands to Launcher methods, and no logic in gcl-ui beyond model adapters. - Error and async discipline: per-module thiserror enums wrapped by a crate-level error, anyhow in binaries, a single tokio runtime owned by Launcher, spawn_blocking for heavy work, and CancellationToken for long tasks. - Vetted dependency table: pinned crate choices for HTTP (reqwest with rustls), hashing, zip, keyring, tracing, clap, Slint, and testing tools like wiremock and insta. - Use Case: Ask the assistant to add a new Modrinth download command to gcl-cli, and it will route through gcl_core::Launcher, reuse the shared HttpClient, instrument with tracing, and return module errors via thiserror instead of panicking. ## Quick Start Read this Skill before writing any Rust code in gcl-core or gcl-cli and follow its layout, error, async, logging, and dependency rules for every change.
