🪄 Human in the Loop

AI-assisted development with humans in control.

A curated prompt library that keeps developers in the loop. Every prompt is designed to enhance your judgment—not replace it. Built on responsible AI principles: transparency, accountability, and human oversight at every step.

• You stay in control - Prompts guide AI analysis, you make the decisions
• Transparent outputs - Clear reasoning, no black-box recommendations
• Quality gates built-in - Security reviews, quality checks
• Learn as you go - Prompts explain the "why," not just the "what"

⚡ Quick Start

# Install globally
npm install -g @human-in-the-loop/cli

# Or use npx (no install needed)
npx @human-in-the-loop/cli search "security review"

🚀 Try it now

# Find what you need
hit search "code review"

# Install a prompt (creates slash command automatically)
hit install prompt/security-review

# Use in Claude Code
/security-review

🧰 What's Inside

📚 21 Prompts

Code Quality & Architecture

• 🔍 code-review-ts - TypeScript review with constructive feedback
• 🔒 security-review - OWASP Top 10, auth flaws, injection detection
• 🏗️ api-design - REST/GraphQL best practices
• 🎯 system-design-review - Architecture & scalability analysis

Testing

• 🧪 unit-test-generator - Generate Jest/Vitest tests with edge cases
• 🎬 e2e-strategy - Playwright/Cypress test planning
• 🥒 bdd-scenarios - Gherkin scenarios from user stories
• 📊 test-coverage-analysis - Coverage gap analysis and risk assessment

CI/CD & Deployment

• ⚡ pipeline-optimization - GitHub Actions cost & speed optimization
• ☁️ aws-deployment-strategy - Lambda, ECS, CDK infrastructure patterns

Documentation

• 📖 api-documentation - Generate API docs from code
• 🗺️ codebase-explainer - Analyze and document repository architecture

Planning & Culture

• 📋 user-story-breakdown - Epic → INVEST stories with acceptance criteria
• 💬 1-on-1-prep - Structured pre-reads for better meetings
• 💙 code-review-empathy - Transform harsh feedback into helpful coaching
• 🔄 team-retrospective - Facilitate effective team retros
• 🎓 learning-path - Personalized learning roadmaps

Governance

• 🛡️ responsible-ai-audit - Audit AI outputs for accuracy, fairness, transparency
• ⚖️ bias-detection - Identify bias in AI-generated content

Meta

• ✨ prompt-optimization - Analyze and improve existing prompts
• 📦 context-pack-builder - Generate framework context packs

🎓 Skills

Skills are persistent context files (not slash commands) that give Claude framework expertise. Copy them to your .claude/ directory.

• ⚡ Angular Modern - Signals, standalone, control flow (16+)
• 🏛️ Angular Legacy - NgModules, RxJS, lifecycle hooks (pre-16)

🔜 Coming Soon

• 🤖 Agent Registry - Reusable autonomous agents
• 🎯 Context Packs - Domain knowledge bundles
• ✅ Evaluators - Output quality validation
• 🛡️ Guardrails - Safety & compliance enforcement

💡 How It Works

What hit install does

1. Copies files to ~/.claude/tools/ (prompt.md + README)
2. Creates a slash command at ~/.claude/commands/{id}.md
3. Registers it in ~/.hit/registry.json for version tracking

Now you can use /security-review directly in Claude Code.

🎨 CLI Commands

hit search [query]              # Find tools
hit install <tool>              # Install + create slash command
hit list                        # Show installed tools
hit update                      # Update CLI to latest version
hit doctor                      # Validate setup
hit contribute <type> <path>    # Submit new tools
hit stats                       # Installation info

Pro tip: Use --no-claude-command to skip slash command creation if you want manual setup.

🛡️ Keeping Humans in the Loop

Every prompt is designed to augment your expertise, not replace it. Here are a few examples:

🗺️ Understand an Unfamiliar Codebase

hit install prompt/codebase-explainer
/codebase-explainer

# Get architecture diagrams, directory breakdowns,
# and a getting-started guide—you decide what to explore next

💙 Give Better Code Review Feedback

hit install prompt/code-review-empathy
/code-review-empathy

# Transform "This is wrong" into "Consider this approach because..."
# Keep the technical substance, improve the delivery

🛡️ Audit AI-Generated Code

hit install prompt/responsible-ai-audit
/responsible-ai-audit

# Before shipping AI output: check accuracy, bias, security, transparency
# You validate, you decide, you ship

See our Responsible AI Playbook for the full framework.

🤝 Contributing

See CONTRIBUTING.md for guidelines on adding prompts, reporting issues, and submitting PRs.

📚 Documentation

• 🚀 Getting Started - Installation & first prompt
• 🛡️ Responsible AI Playbook - Developer-first AI principles
• 🎨 CLI Reference - Complete command guide
• 🏗️ Architecture - System design overview
• 🤝 Contributing - Detailed contribution workflow

🏗️ Tech Stack

• Build: Nx monorepo
• Language: TypeScript (strict mode)
• Package Manager: pnpm
• CLI: Commander.js
• Testing: Jest

📦 Repository Structure

human-in-the-loop/
├── lib/
│   ├── prompts/              # 21 prompts organized by category
│   │   ├── architecture/     # API design, system design review
│   │   ├── ci-cd/            # Pipeline optimization, AWS deployment
│   │   ├── culture/          # 1-on-1 prep, code review empathy, retros
│   │   ├── documentation/    # API docs, codebase explainer
│   │   ├── governance/       # AI audit, bias detection, security review
│   │   ├── mentorship/       # Learning paths
│   │   ├── meta/             # Prompt optimization, context pack builder
│   │   ├── planning/         # User story breakdown
│   │   └── testing/          # Unit tests, E2E, BDD, coverage analysis
│   ├── skills/               # Framework expertise (Angular, NestJS, Nx)
│   ├── agents/               # Coming soon
│   ├── context-packs/        # CI/CD patterns
│   ├── evaluators/           # Coming soon
│   └── guardrails/           # Coming soon
├── src/
│   ├── cli/                  # CLI source code
│   └── governance/           # CI validation scripts
└── docs/                     # Documentation

🎉 What's New

v3.0.0 - Pure XML Prompts & Claude Code Integration

• 🔌 Automatic slash command creation on install
• 🤖 Tool-based workspace analysis (Read/Grep/Glob)
• ✨ Pure XML prompt format for better structure

See CHANGELOG.md for full history.

📄 License

MIT License - see LICENSE

💙 Built with Care

Human-in-the-Loop by codewizwit

Build with care. Ship with purpose.

🐛 Found a bug? Open an issue 💡 Have an idea? Open an issue ⭐ Like what you see? Star the repo!