Improve Loop — Design skill for Claude Code
Run an autoresearch-style keep-or-rollback loop against a measurable target (software only).
How to install Improve Loop
Installs to ~/.claude/skills/r-claude.md-improve-loop/SKILL.md
mkdir -p ~/.claude/skills/r-claude.md-improve-loop && curl -fsSL https://raw.githubusercontent.com/r/claude.md/HEAD/commands/improve-loop.md -o ~/.claude/skills/r-claude.md-improve-loop/SKILL.md Restart Claude Code, or start a new session, for it to be picked up.
What Improve Loop does
description: Run an autoresearch-style keep-or-rollback loop against a measurable target (software only) argument-hint:
Run a Karpathy-style improvement loop on: **$ARGUMENTS**
First read `~/.claude/rules/loops.md` — this command is the executable form of that doctrine. This is a **software-only, git-backed** loop: rollback must be `git reset`. If the target is infra, DNS, deploy, or anything outward-facing
Alternatives in Design
- Polish UI — Polish UI elements by auditing SCSS and TSX markup against the Anytype design system, then fixing discrepancie 8.7k ★
- Sadd Architecture — Software Architecture Design Documents 663 ★
- Prompt UX UI Especialista Es — Entiendo que quieres que cree un prompt para un especialista en UX/UI siguiendo la estructura del ejemplo que 577 ★
Full documentation available on GitHub
View Source RepositoryRelated Skills
UX Evolve
Auto-iterate the lint→polish→re-lint loop on a generated artifact until score crosses 90 or plateaus. Triggers
Design Render Critique Loop
Render a design from its FORGE:DESIGN_SPEC, critique the rendered pixels against the perceptual negatives, and
Screenshot Iterate
Visual loop — implement against a mock, screenshot the running app with a browser tool, compare, and iterate u
Andrew Translate
Andrew Ng translation-agent style three-step translation skill (initial translate, reflect, improve)
Content Brand Voice
Brand voice definition and management command. Creates a brand voice guide to keep tone, style, and personalit
Shiraz Coding Style
Portable TDD-first coding-agent skill for Codex and Claude Code: clarify behavior first, write RED tests, impl
Related Agents
Autoresearch Orchestrator
Use this agent to run an autoresearch experiment session end-to-end — setup, baseline, and a batch of edit→mea
Research Loop
Autonomous iterative research loop — hypothesis, search, ingest, evaluate, keep/discard via checkpoint. Max 3
Infra Engineer.Agent
Use for repository-grounded IaC changes and scoped deployment work. Plan-only by default; apply/rollback is al