✨ This meta-repository collects shared configuration for the organization and serves as a jumping-off point for all other public projects under @jleechanorg. Explore the highlights below to quickly find the automation framework or showcase build that fits your needs.

• 🛠️ AI Development Recommendations - Battle-tested AI coding toolchain
• 🧠 Consensus ML - Multi-model AI consultation platform
• 🎮 WorldAI GenAI RPG - AI-powered tabletop RPG
• 🚀 Flagship Systems
  • 📧 mcp_mail - Multi-agent coordination
  • 🤖 claude-commands - Claude Code automation
    • ⚡ Orchestration Framework
    • 🔄 GitHub PR Automation
  • 🕸️ ai_web_crawler - Web scraping MCP
• 💼 Portfolio Highlights

Evidence-based AI development toolchain recommendations from production usage at WorldArchitect.AI

Based on 19,044 commits over 6 months running a production AI-powered D&D platform. This is NOT a "top 10 AI tools" listicle - it's a battle-tested toolchain extracted from actual production usage, backed by git commit evidence and real workflow patterns.

🔗 View the Recommendations →

• ✅ 19 Tools - Claude Code CLI, pytest, gh CLI, Gemini, beads, tmux, Cursor, Warp, Docker, and more
• ✅ 10 MCP Servers - Multi-agent coordination, testing, Chrome automation
• ✅ 80+ Custom Commands - /copilot, /fixpr, /redgreen, /fake, and specialized workflows
• ✅ Multi-Agent Orchestration - 4 AI agents (copilot, cursor, codex, claude) working autonomously
• ✅ Automatic Quality Gates - /fake hook catches speculative code (15% catch rate, 343 checks/month)
• ✅ 12-Step Development Workflow - Complete lifecycle from planning to deployment

• 📊 19,044 commits analyzed (6 months of production usage)
• 🧪 17.4% test coverage (3,308 testing commits)
• 🤖 7,950 /copilot invocations (PR automation)
• 🔍 343 automatic code quality checks (via /fake hook)
• 💰 $40-60/month (cost-conscious AI development)

• README.md - Executive summary of all 19 tools
• TOOLCHAIN.md - Complete tool inventory with evidence scores
• WORKFLOWS.md - 12-step multi-AI development workflow
• FAKE_HOOK_ANALYSIS.md - Automatic code quality gate analysis
• SETUP_GUIDE.md - Installation instructions

Who This Is For: Developers using Claude Code, Cursor, or other AI coding tools who want to see real production patterns with evidence-based recommendations.

Ask one question, receive synthesized analysis from 5 major AI models (Cerebras, Claude Sonnet 4, Gemini 2.5 Flash, Perplexity Sonar, Grok) with consensus, divergent perspectives, and comprehensive coverage highlighted. Research shows multi-model approaches reduce hallucinations by 60% and outperform single models.

🔗 Explore the Platform →

• Backend: TypeScript FastMCP HTTP/STDIO server + Express middleware
• Frontend: React 19 + TypeScript + Vite with Tailwind CSS
• Conversations: Dedicated MCP server for conversation persistence
• Architecture: "MCP all the way down" - separate services communicate via Model Context Protocol
• Infrastructure: Google Cloud Run, Firebase Auth, Firestore persistence, Secret Manager

• 🤝 2-Round Parallel Architecture - Round 1: All models execute in parallel (~22s), Round 2: Cerebras synthesizes insights (~8s)
• ⚡ Cost Efficiency - Cerebras synthesis at ~$0.60/M tokens vs $5-20/M for premium models
• 📦 Published npm Packages - Reusable @ai-universe/* packages for MCP-based applications
• 🔄 Dual Transport - FastMCP HTTP proxy (production) + STDIO transport (Claude CLI development)
• 📊 Conversation Persistence - MCP-native conversation management with Firestore storage
• 🛡️ Tiered Rate Limiting - Anonymous/authenticated/VIP tiers with distributed coordination

Your Digital D&D 5e Game Master - an AI-powered tabletop RPG platform that generates dynamic narrative content, manages game state, and provides an immersive campaign experience.

🎯 Try the Live Demo →

• Backend: Python 3.11 + Flask 2.x with Google Gemini AI integration
• Database: Firebase Firestore with real-time synchronization
• Frontend: Vanilla JavaScript + Bootstrap 5.x with multi-theme support (light/dark/fantasy/cyberpunk)
• AI Pipeline: Advanced prompt engineering with dual-pass generation, entity tracking, and model fallback

• 🎭 Interactive Story Generation - Real-time AI-powered narrative with planning block enforcement
• ⚔️ D&D 5e Mechanics - Hybrid architecture combining rule-based validation with generative content
• 📊 State Synchronization - Explicit game state tracking with consistency validation
• 📄 Multi-Format Export - Campaign documents in PDF/DOCX/TXT for offline play
• 🎨 Theme System - 5 distinct visual themes with runtime switching
• 🔧 MCP Architecture - Model Context Protocol for clean separation of concerns

• Multi-agent coordination system via async messaging (forked from Dicklesworthstone/mcp_agent_mail).
• 9 core enhancements removing coordination barriers: 65% token reduction via lazy loading, global agent/message lookup across projects, automatic @mention scanning in unified inbox.
• Streamlined workflows: one-step agent registration with auto-project creation, contact-free messaging without approval gates.
• Projects as metadata only (not boundaries), enabling seamless cross-repo agent collaboration with flexible project keys.

Automation backbone that scales Anthropic's Claude Code across heterogeneous environments.

• 150+ purpose-built slash commands compose MCP tools, shell scripts, and remote workflows into repeatable playbooks.
• Pipelines include queueing, rate limiting, observability hooks, and audit-friendly transcripts for full lifecycle coverage.
• Cross-platform setup guidance ensures Linux, macOS, and WSL contributors can bootstrap consistent workspaces in minutes.

Multi-agent orchestration system for complex development tasks using tmux-based agent management:

• Specialized Agents: Frontend, backend, testing, and task-specific agents with complete isolation
• Intelligent Reuse: 50-80% efficiency gains through idle agent detection and strategic reuse
• Real-time Coordination: Redis-backed coordination with progress monitoring via /orch status
• Natural Language Commands: "Build X urgently", "connect to sonnet 1", "monitor agents"
• Branch Isolation Protocol: Agents work in dedicated worktrees preventing context contamination

Seamless CI/CD integration with intelligent PR processing (PyPI: jleechanorg-pr-automation | source):

• Actionable PR Detection: Only processes PRs that need attention based on commit-based tracking
• Safety Limits: Built-in rate limiting (per-PR and global) with cross-process thread-safe persistence
• Hook Integration: Auto-triggers after push operations with full Claude Code workflow integration
• Email Notifications: Optional SMTP integration for automation alerts
• TodoWrite Integration: Tracks automation tasks with comprehensive progress reporting

• Node.js FastMCP façade with Python scraping workers for reliable traversal of the public web.
• Translates natural-language crawl intents into structured jobs that return deduplicated Markdown in real time.
• Built-in sanitizers strip tracking cruft and normalize typography, yielding clean corpora for downstream models.
• Operational resilience via caching, rate limits that respect robots.txt, and resumable streams for transient failures.

🔍 Looking for something else? Explore the full organization at github.com/jleechanorg.