Comprehensive multi-lens code review with presets and a synthesized, prioritized verdict. Use this skill whenever you want existing changes assessed - review PRs, branches, diffs, or sets of changed files with multiple independent lenses (correctness, performance, security, etc).
Structured multi-dimension design exploration that produces a synthesized design doc with options, trade-offs, risks, and a phased plan. Use this skill when you want something thought through before it is built - for API design, data modeling, UX, scalability, security, and integration concerns. Explores problems across independent dimensions and reconciles them.
Enterprise-grade engineering — the shiny workflow with the implementation expanded into the Rule of Five iterative-refinement passes. Use this skill for critical work where quality outweighs speed - financial, payments, auth, billing, or data-migration code where mistakes are serious.
Engineer-in-a-Box — a disciplined design→implement→review→test→submit workflow with gates for any non-trivial code change. Use this skill proactively for feature work, bug fixes, and refactorings that touch multiple files. Follows five phases with gates - design before coding, review before shipping, test before submitting.
Expert knowledge for building web-based GUIs and websites with CLOG (Common Lisp Omnificent GUI), with the declarative `with-clog-create` macro as the recommended way to build UIs. Use this skill whenever the user mentions CLOG, wants to build a Common Lisp web app or browser-rendered desktop-style GUI, asks about initialize/on-new-window/connection-data-item/clog-gui/clog-web/with-clog-create, needs help with WebSocket-driven Lisp UIs, writes a CLOG plugin or new element class, hooks CLOG into a SQL database (clog-dbi / clog-auth), or deploys a CLOG server behind a reverse proxy. Trigger even when the user just says "browser GUI in Common Lisp" or "web UI for my Lisp app" without naming CLOG by name.
Expert knowledge for Asahi Linux — running, troubleshooting, and hacking on Linux for Apple Silicon Macs (M1/M2/M3/M4). Use this skill whenever the user mentions Apple Silicon Linux, Asahi, Fedora Asahi Remix, m1n1, alx.sh, or asks about bootloaders, kernel/GPU/audio/firmware, x86 emulation with muvm/FEX, DFU recovery, or Steam/gaming on M-series Macs — even if they describe the hardware as "M1 MacBook" or "Apple Silicon" without naming Asahi. Also use for contributing upstream (m1n1 hypervisor, kernel drivers, speakersafetyd, Mesa asahi/honeykrisp).
Create and refine autonomous research workflows where an AI agent iteratively experiments on code, measures results, and keeps or discards changes. Use this skill whenever the user wants to set up an autoresearch loop, create a program.md, design an autonomous experimentation workflow, optimize hyperparameters autonomously, run overnight experiments, or adapt an existing autoresearch setup to new metrics, hardware, or goals. Also trigger when the user mentions "autoresearch", "experiment loop", "autonomous optimization", "program.md", or wants an AI agent to try ideas and keep what works.
Expert knowledge for GNU Guix package and service development. Use when creating or modifying Guix packages, system services, home services, or working with a Guix repository checkout. Handles package definitions, build systems, service types, configuration records, and development workflows.