Set up a project for Claude Code without leaking personal config into the project's git history. Creates a per-project overlay under ~/.dotfiles/projects/ (CLAUDE.md, agents/, settings.local.json), symlinks it into the project worktree, and — for compose-based devcontainers — writes a local seed-and-copy override for host SSH/gh/Claude/dotfiles. Also detects and repairs legacy writable or dual-home Claude mounts. Use when starting work on a new project, when an open-source project shouldn't carry your CLAUDE.md, when adding agent-teams support to an existing project, or when the user mentions "set up this project for Claude" / "bootstrap project Claude config" / "share host SSH/gh/dotfiles with the devcontainer".
Review code changes since a commit, classify correctness and maintainability findings, optionally apply only high-confidence safe fixes, and report remaining issues. Use when the user asks to polish changed code, review a commit range, run a pre-PR quality pass, or invokes $my:polish-core.
Holistic codebase review focused on architecture drift, duplicate logic across packages, code smells, quality, tests, and UX — returns up to 10 high-impact improvements (sharp over padded). Reads project conventions from CLAUDE.md/AGENTS.md before the review. Use when the user asks about tech debt, what to refactor, code audit, codebase health, what needs fixing, improvement opportunities, or general code quality review.
Run an autonomous overnight improvement loop on a codebase. Each iteration runs the project's review skill (default `my:improve`), picks the top eligible finding, fixes it, runs project-defined verification gates, and commits or rolls back. After max iterations or no eligible findings, pushes the branch, opens a PR, and runs CodeRabbit autofix rounds. Reads per-project config from `.claude/overnight-config.yaml` (gates, do-nots, iteration cap). Use when the user wants to run an unattended improvement session and wake up to a PR.
Supplementary code review guidance for language-specific idioms, codebase pattern adherence, and complexity analysis. Use this alongside /review to provide deeper, structured feedback beyond what the built-in review covers.
Analyze, re-engineer, or bootstrap projects to align with AI-first design principles. Use when asked to review, audit, improve, 'ai-firstify', or start a new project. Performs deep analysis across 7 dimensions, actively restructures existing projects, or guides new project setup through discovery questions. Based on the 9 design principles and 7 design patterns from the TechWolf AI-First Bootcamp.
Simplify recently modified code for clarity, consistency, and maintainability without changing behavior. Detects languages from changed files and applies language-specific simplification rules alongside general principles.
Check tool availability for the current session. Auto-detects required tools from project context or accepts an explicit list. Groups results by status and suggests install commands for missing tools.