Structured reasoning systems for Claude Code. Instead of prompting from scratch every time, you invoke a system — it runs a defined workflow, enforces reasoning structure, and produces consistent outputs.
Each system is a .md file. Copy it to ~/.claude/commands/ once — then invoke it by name in any Claude Code session.
Live Systems
Takes a product problem — vague idea, named feature, or draft PRD — and moves it through a 16-phase cycle to execution-ready output. The system doesn't agree with your framing. It challenges assumptions, generates competing hypotheses, stress-tests solutions, detects gaps, heals itself, learns from your input, and repeats until all critical gaps are resolved.
Includes a self-evolution mode — the system turns inward, detects drift, and fixes its own logic. And a self-learning mode — after every run, it asks what it got wrong and converts your answers into permanent improvements.
Evaluates a portfolio the way a hiring manager actually reads it — not "looks good," but a structured pass across positioning, content signal density, UX, mobile, and copy quality. Runs a 7-step audit and produces exact before/after rewrites. Includes a hiring simulation that gives you a verdict from four perspectives: Recruiter, Hiring Manager, Design Leader, and AI/Product Leader.
Modular skills can be run independently — hire simulation only, mobile QA only, full-site consistency scan, or execution mode that applies all fixes directly to files.
system-name.md skill file, optional modular skills in skills/, optional named workflows in workflows/. Add a card here with the same structure — title, description, commands, install, download.