omer-faruk-g

ClaudeMT — Code Quality skill for Claude Code

Code Quality community

ClaudeMT v3: Claude Code skill for minimalist architecture planning and diff-only code review.

How to install ClaudeMT

This entry records only its repository, not the path inside it, so there is no exact command to give. Open omer-faruk-g/ClaudeMT and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What ClaudeMT does

ClaudeMT v3: Claude Code skill for minimalist architecture planning and diff-only code review. Prompt-only, one SKILL.md.

Alternatives in Code Quality

Open Code Review — /open-code-review 21.5k ★
Code Quality Review — Review code quality in: $ARGUMENTS 5.8k ★
Ars Lit Review — ARS academic-paper lit-review mode — annotated bibliography in paper format 4.5k ★

README

ClaudeMT

ClaudeMT v3 (Claude Minimalist Thinking & Diff-Reviewer) is a planning and review skill for [Claude Code](https://code.claude.com). It is prompt-only: one `SKILL.md`, with no scripts and no hooks. When you ask for a design, the model answers with bullets, diagrams and a decision matrix, and writes no code until you confirm. When you ask for a review, it reads only the diff and reports only real problems. It does not narrate the code or refactor anything outside the diff.

Rules

Part	Rule	What the model does
1	No verbal brainstorming	No inner monologue, reasoning traces or pros/cons essays. Each argument is stated once.
1	No over-engineering	It designs only for the current scope. No microservices, caching layers or abstractions for future needs you did not ask for.
1	No cascade redesign	When one data structure or interface changes, it adapts only the call sites that change actually breaks.
1	Zero-code gate	A design or planning request gets a symbolic design only. Code comes after you confirm.
1	Symbolic format	Bullets of at most 12 words, Mermaid diagrams or pseudocode, and a fixed decision matrix (option, complexity, memory/latency, trade-off) for every architectural choice.
2	Diff-only scanning	In reviews and refactoring requests it reads only the diff and the changed signatures. It opens or asks for other files only when a breaking interface risk is verified.
2	Surgical impact mapping	Feedback covers only the modified parts: changed parameters, return types and logic branches.
2	No narration	It never walks through what the code does. It reports flaws, breaking risks and efficiency issues.
2	Plausible edge cases only	It skips astronomically unlikely cases and reports practical memory leaks, race conditions, type mismatches and boundary bugs.
2	Review ≠ refactor	A review never turns into a refactor. Fix suggestions stay inside the diff.
2	Review