🚀 AI Command Library

Stop writing the same AI prompts over and over.

Get proven commands for code reviews, PR management, and development workflows that work across Claude, Cursor, Copilot, and Codex.

# Install once, works everywhere
npx @mkaczkowski/ai-command-library -p claude

✅ One install → Works everywhere ⚡ Ready-to-use workflows 🔄 Auto-sync across tools 📚 Battle-tested commands

Why This Matters

Every developer wastes time on:

• 🔄 Writing similar prompts for code reviews and PR management
• 🎯 Getting inconsistent AI results across the team
• ⏰ Recreating workflows across projects
• 🤝 Unable to share effective AI commands

This library fixes all of that.

You get pre-built workflows, consistent setup, team collaboration, and instant productivity.

Quick Start

Get your first AI command working in 2 minutes:

# For Claude Desktop/Code
npx @mkaczkowski/ai-command-library -p claude

# For Cursor IDE
npx @mkaczkowski/ai-command-library -p cursor

# For GitHub Copilot
npx @mkaczkowski/ai-command-library -p copilot

# For Codex CLI
npx @mkaczkowski/ai-command-library -p codex-global

Try it now:

"Use the PR review enhancement workflow to improve my code review comments"

🎉 Done! Your AI tools now have professional workflows.

How It Works

Think of this as shared AI prompts organized into plugins. Instead of writing your own, you get:

plugins/pr/              → PR workflows + scripts
plugins/debugger/        → Debugging tools
plugins/jira/            → JIRA integration
       ↓
  link-ai-commands       → Smart sync
       ↓
.claude/commands/        → Commands
.claude/skills/          → Skills (Claude only)
.claude/agents/          → Agents (Claude only)

Core concepts:

• 📦 Plugins - Grouped commands for specific tasks (PR, debugging, JIRA)
• 📝 Commands - Step-by-step AI instructions
• 💡 Skills - Reusable capabilities (Claude only)
• 🎯 Agents - Specialized subagents (Claude only)
• 🔗 Linking - Auto-sync to your AI tools

Installation Options

Direct Usage (No Install)

npx @mkaczkowski/ai-command-library -p <provider>

Install & Link

npm install -D @mkaczkowski/ai-command-library
npx link-ai-commands --provider <provider>

Auto-Sync on Install

{
  "scripts": {
    "postinstall": "link-ai-commands --provider claude"
  }
}

Choose Your Provider

| Provider | Destination | Skills | Agents | Best For | | --------------- | ------------------- | ------ | ------ | ----------------------- | | claude | .claude/ | ✅ | ✅ | Project-specific Claude | | claude-global | ~/.claude/ | ✅ | ✅ | User-wide Claude | | cursor | .cursor/ | ❌ | ❌ | Cursor IDE | | cursor-global | ~/.cursor/ | ❌ | ❌ | User-wide Cursor | | copilot | .github/prompts/ | ❌ | ❌ | Copilot Workspace | | codex-global | ~/.codex/prompts/ | ❌ | ❌ | Codex CLI |

Provider-specific notes:

• Copilot: Prompts are flattened to hyphenated names ending in .prompt.md
• Codex: Commands are flattened to single filenames (e.g., pr__enhance-review.md)
• Skills/Agents: Only available for Claude providers (project and global)

Multi-tool setup:

# Use multiple AI tools? Run once per provider
npx link-ai-commands --provider claude
npx link-ai-commands --provider cursor

Advanced Options

Install specific plugins:

npx link-ai-commands --provider claude --plugins pr,debugger

List available plugins:

npx link-ai-commands --list-groups

Use symlinks for live updates:

npx link-ai-commands --provider claude --mode symlink

Preview changes:

npx link-ai-commands --provider claude --dry-run

Available Workflows

Ready-to-use workflows that solve real problems. Each includes commands + helper scripts for GitHub integration.

🔧 PR Review Enhancement

Transform blunt code review feedback into collaborative, professional guidance.

Command: pr/enhance-review.md

When to use: After writing initial review comments but before posting them.

# Step 1: Fetch and rewrite comments
node {{script:pr/scripts/fetch-pr-comments.js}} --pr=123
# Then ask AI: "Enhance my review comments"
# Output: tmp/pr-[PR_NUMBER]-comments.md

# Step 2: Update GitHub
node {{script:pr/scripts/edit-pr-comments.js}} --pr=123

Benefits: Professional tone, same technical accuracy, stronger team relationships.

flowchart TD
    Fetch[Fetch PR comments]
    Fetch --> Context[Review diff and project standards]
    Context --> Rewrite[Rewrite in collaborative voice]
    Rewrite --> Output[Save improved feedback]
    Output --> Decide{Push to GitHub?}
    Decide -->|Yes| Update[Update comments]
    Decide -->|No| Share[Manual review]

✍️ Create PR Reviews

AI-powered systematic code review with comprehensive issue detection.

Command: pr/draft-review.md

When to use: When you need to review a new PR thoroughly.

# Step 1: Prepare review findings
node {{script:pr/scripts/fetch-pr-context.js}} --pr=123
# Then ask AI: "Prepare a review for this PR"
# Output: tmp/pr-[PR_NUMBER]-findings.md

# Step 2: Create and post review
# AI generates: tmp/pr-[PR_NUMBER]-review-comments.json
node {{script:pr/scripts/create-pr-review.js}} --input=tmp/pr-123-review-comments.json --pr=123

Benefits: Catch issues you might miss, professional comments, faster reviews.

flowchart TD
    Context[Fetch PR context]
    Context --> Inspect[Study patches and neighboring code]
    Inspect --> Analyze[Assess risks and strengths]
    Analyze --> Output[Generate structured findings]
    Output --> Craft[Compose inline comments]
    Craft --> Map[Align paths and line numbers]
    Map --> Export[Export to JSON]
    Export --> RunScript[Create pending PR review]

✨ Create Pull Requests

Auto-generate comprehensive PR descriptions from your code changes.

Command: pr/create-pr.md

When to use: When you're ready to create a pull request from your feature branch.

# AI analyzes your changes and populates the PR template
# Ask AI: "Create a PR using the create-pr workflow"

# Or provide context upfront:
# "Create a PR. I added a user search feature with debouncing to reduce API calls."

What the AI does:

1. Reads .github/PULL_REQUEST_TEMPLATE.md
2. Analyzes commits and diffs since diverging from main
3. Extracts tickets from branch name/commits
4. Identifies new dependencies and test coverage
5. Populates template with reviewer-friendly content
6. Creates PR using gh pr create

Benefits: Complete descriptions, faster reviews, consistent quality.

🔗 Create Stacked Pull Requests

Problem: Large features are hard to review in a single PR, and breaking them into dependent PRs manually is time-consuming and error-prone.

Solution: Automated creation of sequential dependent PRs that build on each other, following conventional commit format with proper dependency tracking.

What you get:

• 📚 Work broken into logical, reviewable chunks
• 🔗 Proper dependency chain between PRs
• 🎯 Conventional commit titles with stack position tracking
• ⚡ Faster reviews through smaller, focused PRs

Create Stacked PRs (pr/create-stacked-pr.md)

When to use: When you have a large feature that should be broken into multiple dependent PRs for easier review.

# Ask your AI:
# "Create stacked PRs for feat auth PROJ-123 implement OAuth2 authentication flow"

What happens: Your AI:

1. Analyzes the full scope of work needed
2. Breaks it into logical chunks (foundation → features)
3. Plans the stack with proper dependencies
4. Shows you the plan for approval
5. Creates sequential branches with conventional commit format
6. Generates dependent PRs with proper base branches
7. Provides a summary with all PR links and merge order

What you need to provide:

• Conventional commit type (feat, fix, refactor, etc.)
• Scope/feature area
• Ticket number
• Description of the changes

Output: Series of dependent PRs with proper titles like feat(auth): add token validation 1/3 (PROJ-123)

🔄 Address Review Feedback

Systematically resolve all review comments with tracking and validation.

Command: pr/address-review.md

When to use: When you have unresolved review comments to address.

# Step 1: Plan resolutions
node {{script:pr/scripts/fetch-pr-comments.js}} --pr=123 --reaction=+1 --ignore-outdated --include-diff-hunk
# Then ask AI: "Plan resolutions for this feedback"
# Output: tmp/pr-[PR_NUMBER]-address-plan.md

# Step 2: Implement changes
# AI guides you through each planned change
# Output: tmp/pr-[PR_NUMBER]-address-report.md

# Step 3: Reply to resolved comments
node {{script:pr/scripts/reply-to-comments.js}} --pr=123 --dry-run
node {{script:pr/scripts/reply-to-comments.js}} --pr=123

Benefits: Nothing gets missed, appropriate responses, faster resolution cycles.

flowchart TD
    Gather[Fetch unresolved threads + PR context]
    Gather --> Study[Review diffs and previous discussion]
    Study --> Strategize[Define fixes and validation per comment]
    Strategize --> Document[Write implementation plan]
    Document --> Implement[Apply fixes, docs, and tests]
    Implement --> Validate[Run targeted checks]
    Validate --> Report[Generate final report]
    Report --> Reply[Reply with commit references]
    Reply --> Wrap{Everything resolved?}
    Wrap -->|Yes| Share[Push commits and close loop]
    Wrap -->|No| Loop[Return to remaining feedback]

🐞 File JIRA Tickets

Quick JIRA ticket creation with guided workflows:

Bug Reports (jira/create-bug.template.md)

# Ask AI: "File a JIRA bug for [issue]"

Outputs: Complete description + ready-to-run jira issue create command

Technical Debt (jira/create-tech-debt.template.md)

# Ask AI: "Create a technical debt ticket for [work]"

Outputs: Structured ticket with What/Where/Why + acceptance criteria

Flaky Tests (jira/create-flaky-test.template.md)

# Ask AI: "File a flaky test ticket for [test]"

Outputs: Complete diagnostic details + reproduction steps

Prerequisite: Install jira-cli and authenticate.

Script Integration

Commands reference helper scripts using placeholders that work everywhere:

# This works across all providers:
node {{script:pr/scripts/fetch-pr-comments.js}} --pr=123

# Resolves to correct path for each provider:
# Claude:  .claude/commands/pr/scripts/...
# Cursor:  .cursor/commands/pr/scripts/...
# Copilot: .github/prompts/pr/scripts/...
# Codex:   ~/.codex/prompts/pr/scripts/...

Available placeholders:

• {{script:path/to/script.js}} - References helper scripts
• {{path:commandsRoot}} - References commands directory

Helper Scripts Reference

• fetch-pr-comments.js - Get review comments with filtering
• fetch-pr-context.js - Gather PR metadata, files, commits

node {{script:pr/scripts/fetch-pr-comments.js}} --pr=123 --reaction=+1
node {{script:pr/scripts/fetch-pr-context.js}} --pr=123

• create-pr-review.js - Submit comment JSON as review
• edit-pr-comments.js - Update existing comments
• reply-to-comments.js - Post resolution acknowledgments

node {{script:pr/scripts/create-pr-review.js}} --input=tmp/pr-123-review.json --pr=123
node {{script:pr/scripts/edit-pr-comments.js}} --pr=123
node {{script:pr/scripts/reply-to-comments.js}} --pr=123 --dry-run

💡 All scripts support --help for usage details.

Prerequisites

• Node.js 22+ and npm 10+
• GitHub CLI for PR workflows
  • Run gh auth login with repo scope
  • Enterprise users: set GH_HOST or use --repo flags

Alternative Installation

Install directly from GitHub:

npm install -D git+https://github.com/mkaczkowski/ai-command-library.git

Development & Contributing

Setup

git clone https://github.com/mkaczkowski/ai-command-library.git
cd ai-command-library
npm install

Code Quality

npm run lint           # Check code style
npm run format:check   # Verify formatting
npm run lint:fix       # Fix issues
npm run format         # Apply formatting

Repository Structure

ai-command-library/
├── plugins/              # Plugin groups
│   ├── pr/              # PR workflows
│   ├── debugger/        # Debugging tools
│   ├── jira/            # JIRA integration
│   └── common/          # Shared utilities
├── providers/           # Provider configs
├── scripts/             # Core linking logic
└── bin/                 # CLI entry point

Plugin Organization

Each plugin group can contain:

{group}/
├── commands/      # Markdown command templates + scripts
├── skills/        # Claude Code Skills (Claude only)
└── agents/        # Claude Code Subagents (Claude only)

Available groups:

• pr/ - Pull request workflows and utilities
• debugger/ - Debugging agents and skills
• jira/ - JIRA ticket creation
• common/ - Cross-cutting utilities

Release Process

1. Update package.json version and CHANGELOG.md
2. Run npm run release to verify
3. Publish: npm publish --access public
4. Push release commit and tag

Support & Community

Need Help?

• 🐛 Bug reports: GitHub Issues
• 💡 Feature requests: GitHub Issues
• 🔒 Security concerns: See SECURITY.md

Contributing

We welcome contributions:

• Add new workflows
• Improve existing templates
• Enhance provider support
• Update documentation

License

Released under the MIT License.

Happy coding! 🚀