A curated collection of AI-powered slash commands for code review, problem analysis, and refactoring workflows.

📝 /create-plan - Create comprehensive task plans
📋 /plan-review - Review implementation plans
✅ /finalize-plan - Finalize completed tasks
🔍 /code-review-low - Fast code review
🔬 /code-review-high - Thorough code review
🐛 /problem-analyzer - Identify bugs and affected files
♻️ /refactor-code - Start refactoring workflows
🔌 /kill-port - Kill processes running on specific ports
🔎 /research-better-lib - Find modern, faster library alternatives
📦 /ask-codefetch - Gather codebase context into a single markdown file
🚀 /ask-gpt-pro - Generate codebase, copy to clipboard, and open AI chat
⚛️ /solidjs-rules - Comprehensive SolidJS coding rules
🧹 /component-cleanup - SolidJS component size and structure rules
✂️ /short-tokens - Minimize token usage while retaining key information

Copy the .md files from the slash-commands/ directory into .cursor/commands/ directory. Remove the $1 parameters and XML wrapping tags when using in Cursor.

Official docs: Cursor Slash Commands

Copy the .md files from the slash-commands/ directory into ~/.codex/prompts/ directory. Restart Codex after installation for changes to take effect.

Official docs: Codex Prompts

Claude Code supports both project-specific and personal commands:

Project commands (shared with team):

mkdir -p .claude/commands
# Copy the .md files to .claude/commands/

Personal commands (available across all projects):

mkdir -p ~/.claude/commands
# Copy the .md files to ~/.claude/commands/

Official docs: Claude Code Slash Commands

Creates a comprehensive task plan with structured sections for context, success criteria, deliverables, approach, risks, and testing. Ensures plans follow best practices and include all necessary information before implementation.

Features:

• Structured plan template with all required sections
• Context and problem statement
• Measurable success criteria
• Clear deliverables and approach
• Risk assessment and testing strategy
• Cross-references to AGENTS.md for rules/constraints

Usage: /create-plan TASK="<task-description-or-goal>"

Examples:

/prompts:create-plan TASK="Implement user authentication system"

/prompts:create-plan TASK="Refactor the payment processing module"

Reviews implementation plans and provides a go/no-go decision for development. Evaluates:

• Codebase alignment and current patterns
• Scope clarity and completeness
• Performance, security, and privacy impact
• Code smells and potential caveats

Usage: /plan-review "<your-plan>"

Examples:

/plan-review "We're planning to implement plan @task-x-new-cool-feature.md"

/plan-review "Here is the plan we want to implement: @my-plan.md"

Finalizes a completed task by creating two artifacts: a finished plan file and a lightweight summary for fast AI ingestion. Maintains traceability while keeping summaries token-friendly.

Features:

• Moves task to finished directory with status
• Creates lightweight summary file (SOW - Summary of Work)
• Cross-references between plan and summary
• Tracks affected files, tests, and decisions
• Token-optimized for AI consumption

Usage: /finalize-plan TASK_FILE="<path-to-task.md>"

Examples:

/prompts:finalize-plan TASK_FILE="docs/tasks/todo/01-user-auth.md"

/prompts:finalize-plan TASK_FILE="docs/tasks/todo/05-payment-refactor.md"

Comprehensive code review commands for post-implementation analysis.

code-review-low: Fast review focusing on critical issues

• Code smells and security vulnerabilities
• Performance bottlenecks
• Essential test coverage

code-review-high: Thorough review with extended analysis

• All code-review-low checks plus additional validations
• Deeper security and performance analysis
• Comprehensive test recommendations

Usage: /code-review-low "<files/code/description to review>"

Examples:

/code-review-low "my plan @task-x-cool-feature has been implemented check unstaged files for review"

/code-review-low "@auth.server.ts @auth.rs @auth.connector.ts"

Identifies bugs and maps affected files across the codebase. Provides:

• Root cause analysis
• File impact assessment
• Minimal safe fix proposals
• Documentation gap identification

Usage: /problem-analyzer "<problem-description>"

Examples: The Problem analyzer prompt works best when you add logs and a short problem description.

I ran this often with cheeta model to get a quick analysis and move over to gpt codex.

/problem-analyzer "the app won't compile. here are the logs <paste-logs>"

/problem-analyzer "we have a serious blocking bug in @user-form.ts and @user.service.ts"

Initiates refactoring workflows with clear scope and isolation:

• Feature-specific refactoring
• Commit isolation guidelines
• Documentation of unrelated issues (without fixing)
• No backward compatibility or feature flags

Usage: /refactor-code "<refactoring-goal>"

/refactor-code "here is our implementation plan @impl-plan lets start"

Kills processes running on specified ports with OS-specific commands and safety checks:

• Cross-platform support (macOS, Linux, Windows)
• Process identification and verification
• Safety warnings for system processes
• Graceful shutdown recommendations

Usage: /kill-port "<port-number(s)>"

Examples:

/kill-port "3000"

/kill-port "3000 8080 9000"

A comprehensive template for evaluating alternatives to baseline libraries. Helps find modern, faster JavaScript/TypeScript libraries that outperform existing solutions in latency and bundle size while maintaining or improving relevance/quality.

Features:

• Structured problem statement and success metrics
• Search query templates for benchmarking
• Candidate library shortlist evaluation
• Minimal benchmark design guidelines
• Decision rubric and deliverables checklist

Usage: /research-better-lib "<baseline-library> <domain/use-case> <candidate-libraries>"

Examples:

/research-better-lib "Fuse.js fuzzy search fuzzysort quick-score FlexSearch MiniSearch"

/research-better-lib "lodash utility functions ramda remeda"

Gather narrowed codebase context using codefetch into a single markdown file. Research relevant files/directories, generate scoped context, and use the output file as context.

Workflow:

1. Research: identify relevant files/directories for the task
2. Generate: run codefetch with scoped files → single .md file
3. Read: use codefetch/codebase.md as context (not individual files)

Features:

• Single consolidated markdown file output
• Scoped file inclusion with globs and directories
• Token limits and encoding options
• Project tree visualization
• Extension filtering and exclusion patterns

Usage: /ask-codefetch [PROMPT="<task-description>"]

Examples:

# Generate context file
npx codefetch --max-tokens 50000 \
  --output codefetch/codebase.md \
  --token-encoder o200k \
  --project-tree 3 \
  --include-files "src/auth/**/*.ts,src/api/**/*.ts" \
  --include-dir "src/utils"

Generate codebase with codefetch, copy to clipboard, and automatically open AI chat. Streamlined workflow for quick codebase analysis.

Workflow:

1. Research: identify relevant files/directories for the task
2. Run: codefetch open with scoped files → copies to clipboard, opens browser
3. Paste: codebase in clipboard, paste into AI chat

Features:

• Automatic clipboard copy
• Browser opening with AI chat (ChatGPT, Gemini, Claude)
• Custom chat URLs and models
• Copy-only mode (no browser)
• Same scoping options as ask-codefetch

Usage: /ask-codefetch-pro [PROMPT="<task-description>"]

Examples:

# ChatGPT (default)
npx codefetch open --max-tokens 50000 \
  --token-encoder o200k \
  --project-tree 3 \
  --include-files "src/**/*.ts"

# Gemini
npx codefetch open --chat-url gemini.google.com --chat-model gemini-3.0 \
  --max-tokens 50000 --include-files "src/**/*.ts"

# Claude
npx codefetch open --chat-url claude.ai --chat-model claude-3.5-sonnet \
  --max-tokens 50000 --include-files "src/**/*.ts"

# Copy only (no browser)
npx codefetch open --no-browser --max-tokens 50000 --include-files "src/**/*.ts"

Comprehensive SolidJS coding rules and best practices covering reactivity, control flow, async patterns, props, styling, file layout, and tooling.

Features:

• Reactivity basics and signal usage
• Control flow helpers (Show, For, Switch)
• Async and side effect patterns
• Component API and prop guidelines
• Styling conventions
• File and folder organization
• ESLint configuration examples

Usage: /solidjs-rules CODEFILES="<code-files-to-review>"

Examples:

/prompts:solidjs-rules CODEFILES="@Button.tsx @TodoList.tsx"

/prompts:solidjs-rules CODEFILES="src/components/**/*.tsx"

SolidJS component size and structure rules for code cleanup. Helps maintain clean, maintainable components by enforcing size limits and structure guidelines.

Features:

• Component size guidelines (150-250 LOC target, 600 LOC hard cap)
• File naming conventions
• Component splitting strategies
• Logic extraction to hooks/utilities
• JSX nesting limits

Usage: /component-cleanup CODEFILES="<component-files-to-review>"

Examples:

/prompts:component-cleanup CODEFILES="@UserProfile.tsx"

/prompts:component-cleanup CODEFILES="src/components/**/*.tsx"

Minimizes token usage in text while retaining all key information. Aggressively trims text for token savings by removing redundancy and filler, while preserving every key fact, figure, and conclusion. Avoids asterisk formatting to save tokens.

Key principles:

• Trim aggressively for token savings
• Retain every key fact, figure, and conclusion
• Remove redundancy and filler
• Avoid ** asterisk formatting
• Every token consumes space in agent token windows

Usage: /short-tokens TEXT="<text-to-shorten>"

Examples:

/prompts:short-tokens TEXT="The authentication system requires users to provide their credentials, including both a username and a password. This is very important for security purposes."

/prompts:short-tokens TEXT="@long-documentation.md"

• Use create-plan at the start of new tasks to ensure comprehensive planning
• Use plan-review before starting complex implementations
• Use finalize-plan when tasks are completed to maintain organized documentation
• Use code-review-low after implementing features for quick validation
• Use code-review-high for major changes or before releases
• Combine problem-analyzer with debugging sessions for faster issue resolution
• Use ask-codefetch to gather codebase context into a single markdown file
• Use ask-codefetch-pro for quick codebase analysis with automatic clipboard and browser opening
• Use solidjs-rules and component-cleanup when working with SolidJS codebases
• Use short-tokens to optimize text content for AI token windows

• X/Twitter: @kregenrek
• Bluesky: @kevinkern.dev
• Website: Learn to build software with AI: instructa.ai

• OpenAI Codex Tools codex-1up
• AI Prompts - Curated AI Prompts for Cursor AI, Cline, Windsurf and Github Copilot
• codefetch - Turn code into Markdown for LLMs with one simple terminal command
• aidex A CLI tool that provides detailed information about AI language models, helping developers choose the right model for their needs.# tool-starter