lobby-ai 1.1.1

🏢 LOBBY - AI concierge service that multiplies your existing CLI tools

🏢 LOBBY - AI Concierge for CLI Tools

Multiply your existing CLI tools with intelligent AI orchestration

LOBBY is an AI concierge service that works with your existing CLI tools (Claude, Gemini, Cursor, etc.) to provide intelligent task routing and execution planning. It doesn't replace your tools - it makes them smarter.

⚡ Quick Install

# Node (npx wrapper)
npx lobby-ai --help

# Python (user local)
pip install lobby-ai

# Isolated, recommended for CLIs
pipx install lobby-ai
# or
uvx lobby-ai --help

That's it! 🎉

🚀 What is LOBBY?

LOBBY multiplies your CLI tools by providing:

1. 🎯 Intelligent Routing - Optimal model selection for each task type
2. 💰 Cost Optimization - Prefers free models, shows real costs
3. 🔌 MCP Server - Works with Claude CLI, Gemini CLI, Cursor, any MCP tool
4. 🏢 NYC Concierge - Professional service that elevates your workflow

🛠️ How It Works

LOBBY runs as an MCP (Model Context Protocol) server that your existing CLI tools can connect to:

• Claude CLI → calls LOBBY for smart orchestration
• Gemini CLI → calls LOBBY for cost optimization
• Cursor IDE → calls LOBBY for task routing
• Any MCP tool → gets LOBBY's AI concierge services

🎯 What You Get

Intelligent Routing

• Coding tasks → DeepCoder (free) or Claude-3.5 Sonnet
• Writing tasks → GPT-4 or Claude-3 Opus
• Analysis tasks → Gemini 1.5 Pro or Llama-3.1
• Creative tasks → DALL-E + GPT-4 or Midjourney API

Cost Optimization

• Prefers FREE models (agentica-org/deepcoder:free, etc.)
• Routes to cheapest option that meets quality requirements
• Shows real-time cost estimates
• Usage tracking and billing

Professional Execution Plans

• Multi-step actionable plans with shell commands
• Quality assurance checkpoints
• Expected deliverables and outcomes

🚀 Setup (30 seconds)

# 1. Install with basic features
pip install lobby-ai

# Or install with interactive prompts (recommended)
pip install 'lobby-ai[interactive]'

# 2. Configure (auto-detects Claude, Gemini, etc.)
lobby setup

# 3. Start using through your existing CLI tools!

✨ Interactive Setup Wizard

With the [interactive] extras installed, you get a beautiful setup wizard:

lobby setup
# ✅ Interactive API key entry (secure password input)
# ✅ Choose default agent from menu (developer, writer, analyst, etc.)
# ✅ Select preferred AI model with cost indicators
# ✅ Configure tools and preferences

📋 Usage Examples

Interactive Request Mode

# Interactive mode with menus and selections
lobby request "Build a web scraper"
# ✅ Select agent specialization
# ✅ Choose AI model (with FREE options highlighted)
# ✅ Pick tools to enable
# ✅ Configure advanced options

# Non-interactive for automation
lobby --no-interactive request "Build a web scraper" \
  --agent developer --model auto --tools code-execution,web-search

Pattern Management

# Create reusable task patterns
lobby patterns new
# ✅ Interactive pattern builder with variables
# ✅ Save templates for common tasks

# Run saved patterns
lobby patterns run "Code Review"
# ✅ Fill in template variables
# ✅ Execute with saved preferences

# List all patterns
lobby patterns list

In Claude CLI:

claude> Use LOBBY to build a Python web scraper for Hacker News
claude> Ask LOBBY to analyze my GitHub repo structure
claude> Get LOBBY to write a technical blog post about AI agents

In Gemini CLI:

gemini> Use LOBBY orchestration to create a React dashboard
gemini> Have LOBBY optimize my database queries

🏢 MCP Tools Available

Once installed, these tools are available to all your CLI clients:

• orchestrate_task - Full AI task orchestration with optimal routing
• analyze_routing - Preview which models would be selected
• check_usage - View your usage and billing status

💳 Pricing

• Free: 10 orchestrations per day
• After free tier: $0.01 per orchestration
• Professional: $29/month (2,500 orchestrations)
• Enterprise: $99/month (unlimited)

Note: You still pay model providers directly (OpenRouter, Anthropic, etc.). LOBBY charges only for the orchestration service.

⚙️ Requirements

• Python 3.8+
• An OpenRouter API key (get free credits at openrouter.ai)
• At least one MCP-compatible CLI tool:
  • Claude CLI
  • Gemini CLI
  • Cursor IDE
  • Any other MCP-compatible tool

🎯 Philosophy

LOBBY multiplies your existing tools instead of replacing them.

We believe the future is multi-model, multi-provider. Instead of being locked into one AI company's tools, LOBBY lets you:

✅ Keep using Claude CLI, Gemini CLI, Cursor, etc.
✅ Add intelligence through optimal model routing
✅ Save money with cost optimization and free models
✅ Get better results with task-specific model selection

📚 Further docs

• Build from source (pyproject):

  • python -m build # requires build
  • pip install -U build && python -m build
  • pip install dist/*.whl

• Pre-commit hooks: pre-commit install

• Lint/format: ruff format . && ruff check . --fix

• See USER_GUIDE.md for a complete command reference and user journey walkthrough

• See IMPLEMENTATION_SUMMARY.md for architecture details and testing

📚 Support

• Website: lobby.directory
• GitHub Issues: github.com/franco/lobby
• Email Support: [e-mail]

License

MIT License - see LICENSE file.

---

Made with ❤️ in NYC - Your AI concierge is ready to serve!