ai-skills

Manage the full WebAssembly component lifecycle with the `component` CLI tool — init projects, install dependencies, build, run, compose, publish, pull, search, and inspect components against OCI registries (GHCR, Docker Hub, ACR) and meta-registries. Covers the WebAssembly Component Model and WASI Preview 2.

Build a single WebAssembly component in multiple languages (Rust, Python, JavaScript, Go) from a single prompt. Generates parallel implementations with language-suffixed directories so you can compare output, size, and behavior across toolchains. Defaults match wasm-build — Rust + Python on WASI 0.3 RC, JavaScript + Go on WASI 0.2.

Build WebAssembly components from Rust, Python, JavaScript/TypeScript, or Go using the Component Model. Defaults to WASI 0.3 (RC) for Rust and Python (native async / stream / future), and WASI 0.2 for JavaScript and Go (whose toolchains don't yet ship a p3 path). Covers scaffolding, language-specific toolchain setup (wit-bindgen, componentize-py, jco / componentize-js, tinygo), compilation, WIT binding generation, validation, and size/startup optimization.

Upstream Bytecode Alliance utilities for WebAssembly — `wkg` (OCI push/pull with full annotation control, WIT-package authoring via `wkg wit fetch/build/publish`, GHCR/Docker auth) and `wasm-tools` (validate, inspect, embed, extract, compose, print, dump, metadata). Includes a curated catalog of useful pre-built components for discovery.

Run, debug, and profile WebAssembly components and core modules with the `wasmtime` runtime. Covers `run`, HTTP `serve`, `--invoke` with WAVE syntax for direct export calls, AOT `compile` to .cwasm, `wizer` snapshotting, and the inspection commands `objdump` / `explore` / `settings`. Supports WASIp2 and WASIp3.

Use this skill when working with just command runner, Justfiles, or build recipes. Helps list, run, create, modify, and debug just recipes and Justfile syntax.

Use this skill when running untrusted code in hardware-isolated sandboxes, registering host tools for guest code, working with Hyperlight micro-VMs, or using the hyperlight-sandbox Python SDK. Covers sandbox creation, code execution, tool dispatch, file I/O, network allowlisting, and snapshot/restore.