Cursor Command Suite

A set of Cursor commands designed to help developers work faster, onboard quicker, and ship better code using agentic AI. Each command has a single responsibility and produces a markdown file as its output so context is always persistent across sessions.

Check out the NPM package cursor-suite for an easy installation that automatically sets up all commands and rules in your project.

---

How It Works

Commands are organized into three tracks. Each track covers a different phase of the development lifecycle.

cursor-suite/
├── README.md
├── .cursor/
│   ├── commands/           ← .md files = real slash commands (type / to invoke)
│   │   ├── analyze.md
│   │   ├── fix-todo.md
│   │   ├── todo-security.md
│   │   ├── todo-tests.md
│   │   ├── todo-quality.md
│   │   ├── todo-improvements.md
│   │   ├── planner.md
│   │   ├── executor.md
│   │   ├── pr-prep.md
│   │   ├── onboard.md
│   │   └── explain.md
│   └── rules/              ← .mdc files = behavior rules (reinforce each command)
│       ├── analyze.mdc
│       ├── fix-todo.mdc
│       └── ...
└── docs/
    ├── commands/
    │   ├── README.md       ← setup instructions and command index
    │   └── *.md            ← detailed doc for each command
    ├── rules/
    │   ├── README.md       ← how rules work and how to write them
    │   └── *.mdc           ← rule source files
    └── mcp/
        └── README.md       ← MCP setup configs

Two files per command:

• .cursor/commands/analyze.md — the prompt that runs when you type /analyze
• .cursor/rules/analyze.mdc — the behavior rules that enforce how it should act

Commands work without the rules, but the rules make them significantly more reliable.

---

Getting Started

Option A — Install via npm (recommended)

From the root of any project:

# Install
npm i cursor-suite

# Initialize commands and rules
npx cursor-suite init

This automatically creates .cursor/commands/ and .cursor/rules/ and copies all files into them. Safe to re-run — existing files are never overwritten.

Option B — Install manually

mkdir -p .cursor/commands
cp /path/to/cursor-config/docs/commands/*.md .cursor/commands/

mkdir -p .cursor/rules
cp /path/to/cursor-config/docs/rules/*.mdc .cursor/rules/

See docs/commands/README.md for full setup details.

---

Step 1 — Install Repomix

The /analyze and /onboard commands require Repomix to pack your codebase into AI-friendly context:

npm install -g repomix
# or
brew install repomix

Step 2 — Open Cursor in Agent mode and type /

/analyze          → full codebase audit
/onboard          → developer onboarding guide
/planner          → implementation plan for a feature

---

Tracks

🔍 Track 1 — Audit & Fix

Start with /analyze then use /fix-todo or the scoped variants to execute fixes.

Command	Output File	Description
/analyze	repo-breakdown.md	Full codebase audit covering tests, vulnerabilities, code quality, and improvements
/fix-todo	updates repo-breakdown.md	Works through every item in the audit checklist in priority order
/todo-security	updates repo-breakdown.md	Scoped pass — 🔴 security items only, CRITICAL → HIGH → MEDIUM → LOW
/todo-tests	updates repo-breakdown.md	Scoped pass — 🧪 missing and partial tests only
/todo-quality	updates repo-breakdown.md	Scoped pass — 🔧 code quality items only
/todo-improvements	updates repo-breakdown.md	Scoped pass — 💡 improvement items only

Typical flow:

/analyze → review repo-breakdown.md → /todo-security → /todo-tests → /fix-todo

---

🏗️ Track 2 — Feature Development

The Planner thinks and documents. The Executor implements. They communicate through a shared markdown file.

Command	Output File	Description
/planner	docs/implementation-plan/{task-slug}.md	Breaks down a feature into a step-by-step implementation plan
/executor	updates docs/implementation-plan/{task-slug}.md	Executes one task at a time from the plan
/pr-prep	pr-prep.md	Generates a PR description summarizing what changed, why, and what to test

Typical flow:

/planner [describe feature] → review plan → /executor → verify → /executor → /pr-prep

---

🧭 Track 3 — Onboarding & Understanding

Command	Output File	Description
/onboard	onboard.md	Plain English guide to a codebase for a developer joining the project
/explain	explain.md	Deep dive on a specific file or function — what it does, why it exists, how it connects

Typical flow:

/onboard → read onboard.md → /explain [specific file] → /planner [start building]

---

Shared Conventions

Every command produces a markdown file. Context is never lost between sessions. An Executor can pick up where it left off by reading the implementation plan. A new developer can read onboard.md before their first standup.

Workflows never do more than one thing. analyze only audits. fix-todo only fixes. This makes it clear which command to reach for and easy to trust what it will and won't touch.

The Planner never writes code. The Executor never changes the plan. This separation prevents the most common failure mode of agentic AI — executing before thinking, or overthinking instead of executing.

TDD is enforced for all new code. The Executor writes failing tests before writing implementation. The test must fail first to prove it's real.

---

Docs

Section	Description
docs/commands/README.md	How to set up commands, how slash commands work, and how to add new ones
docs/rules/README.md	How rules work, how to install them, and how to write new ones
docs/mcp/README.md	MCP setup configs with credentials guidance for every recommended server

---

Command Reference

Command	Doc
/analyze	docs/commands/analyze.md
/fix-todo	docs/commands/fix-todo.md
/todo-security	docs/commands/todo-security.md
/todo-tests	docs/commands/todo-tests.md
/todo-quality	docs/commands/todo-quality.md
/todo-improvements	docs/commands/todo-improvements.md
/planner	docs/commands/planner.md
/executor	docs/commands/executor.md
/pr-prep	docs/commands/pr-prep.md
/onboard	docs/commands/onboard.md
/explain	docs/commands/explain.md