Skip to main content
Glama
H0wZy

@h0wzy/mcp

by H0wZy
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Go
Local

🚀 H0wZy/mcp — The Ultimate Multi-Agent MCP Hub & Go CLI

CI Pipeline GitHub Release npm Go Version Node Version License: MIT Platform

Centralize, enhance, and distribute high-performance MCP (Model Context Protocol) servers connecting the world's leading AI developer CLIs:

  • Claude Code (Anthropic)

  • OpenAI Codex CLI (OpenAI GPT-5.6 / GPT-6 Astra)

  • Google Antigravity (Gemini 3.1 Pro / Flash)

Cross-model code reviews, independent second opinions, and autonomous multi-agent validation — configured effortlessly via an interactive Golang TUI CLI and distributed via standalone binaries and npx @h0wzy/mcp.

🏷️ #mcp #ai-agents #multi-agent #antigravity #claude-code #openai-codex #gemini #cli #go #bubbletea #tui #developer-tools #model-context-protocol

🏛️ Interactive Architecture Diagram

Explore the full interactive system architecture with Light/Dark themes and semantic tracing: 👉 View Interactive Architecture Diagram

Related MCP server: Multi-MCP

🌟 Why H0wZy/mcp?

  1. DRY Shared Core (@h0wzy/mcp-shared):

    • Eliminates matrix code duplication across host agents. A single createMcpServer() engine manages 100% of JSON-RPC 2.0 stdio protocol handling, error catching, and lifecycle handshakes.

  2. Native Cross-Platform (Zero Windows Glitches):

    • Eliminates POSIX-only bugs like PATH.split(':') by utilizing native path delimiters (; on Windows, : on POSIX).

    • Resolves executable extensions automatically (.exe, .cmd, .bat from PATHEXT), finding agy.exe and codex.cmd without requiring manual environment path overrides.

  3. Instant Performance (Zero npx Latency):

    • Directly executes local binaries with sub-50ms invocation overhead, removing runtime npx network/cache-checking delays.

  4. Resilient Rate Limits & Quotas:

    • Gracefully traps HTTP 429 and ResourceExhausted provider quota limits, returning structured { isError: true } responses so host agents fall back seamlessly without crashing the session.

  5. Interactive Go CLI (h0wzy-mcp):

    • Auto-detects local CLI installations (claude, codex, agy), tests binary health, and guides the user through an interactive checklist to configure MCP connections globally or per-project.

📁 Repository Layout

H0wZy/mcp/
├── .github/workflows/          # CI/CD & cross-platform testing
├── cli/                        # Interactive CLI in Go (Bubble Tea / Huh)
│   ├── cmd/                    # Commands: install, doctor, list, remove
│   ├── detector/               # Discovers local claude, codex, and agy CLIs
│   ├── config/                 # Read/write ~/.claude.json, config.toml, etc.
│   └── ui/                     # Terminal user interface
├── servers/                    # Decoupled, host-agnostic MCP servers
│   ├── antigravity/            # Google Antigravity bridge (Gemini 3.1 Pro / Flash)
│   ├── codex/                  # OpenAI Codex CLI bridge (GPT-5.6 / GPT-6 Astra)
│   └── claude/                 # Claude Code bridge
├── shared/                     # Reusable core (@h0wzy/mcp-shared)
│   ├── server.js               # createMcpServer() generic JSON-RPC 2.0 stdio engine
│   ├── executor.js             # Child process runner with timeouts and buffers
│   ├── resolver.js             # Cross-platform executable resolver
│   └── errors.js               # Resilient Quota / HTTP 429 error handler
├── npm/                        # Lightweight npx runner wrapper
├── specs/                      # SpecKit feature specs & Archify diagrams
├── LICENSE                     # MIT License
└── package.json                # Monorepo workspaces configuration

🚀 Quick Start

1. Interactive Setup (Recommended)

Choose your preferred way to run H0wZy/mcp:

# 1. Instant execution via npx (Zero setup):
npx @h0wzy/mcp

# 2. Or install globally via npm (provides 'hmcp', 'hwzmcp', and 'h0wzy-mcp'):
npm install -g @h0wzy/mcp
hmcp

# 3. Or directly from source with Go:
go run ./cli setup-path
hmcp

📦 Precompiled Standalone Binaries (v1.0.2)

Download zero-dependency native binaries directly from Releases:

The CLI will scan your system:

🚀 H0wZy/mcp — Multi-Agent MCP Hub Setup
Scanning local AI developer CLIs...
  ✓ Claude Code (C:\Users\...\claude.exe)
  ✓ OpenAI Codex CLI (C:\Users\...\codex.cmd)
  ✓ Google Antigravity (C:\Users\...\agy.exe)

? Select MCP bridges to configure:
  [x] Claude Code -> Google Antigravity (Gemini 3.1 Pro/Flash)
  [x] Claude Code -> OpenAI Codex (GPT-5.6 / GPT-6 Astra)
  [ ] OpenAI Codex -> Google Antigravity (Gemini 3.1)

? Configuration Scope: User (Global across all projects)
? Apply configuration now? Yes

✅ All selected bridges configured successfully!

2. Commands & Management

# Check version and verify if updates are available
hmcp version

# Upgrade hmcp to latest release (or 'hmcp update')
hmcp upgrade

# Setup PATH and shims (~/.local/bin) so 'hmcp' works everywhere
hmcp setup-path

# Diagnose local environment, paths, and agent communication health
hmcp doctor

# Diagnose and output as JSON
hmcp doctor --json

# Install all supported integrations automatically
hmcp install --all

# Install a specific bridge globally or locally
hmcp install claude-antigravity --scope user
hmcp install claude-codex --scope project

# List available bridges
hmcp list

# Remove an integration
hmcp remove claude-antigravity

🔌 Available MCP Tools (Symmetrical Parity)

Provider

Tool Name

Description

Antigravity

ask_antigravity

Independent second opinion or general inquiry from Gemini 3.1 Pro / Flash

Antigravity

review_antigravity

Comprehensive code & security review inspecting correctness, edge cases, and diffs

Antigravity

brainstorm_antigravity

Architectural exploration, trade-offs, and design patterns with Gemini

Antigravity

plan_antigravity

Structured implementation roadmaps and dependency-ordered execution steps

Codex

ask_codex

Cross-verification with OpenAI Codex (GPT-5.6 Terra / GPT-6 Astra)

Codex

review_codex

Structured repository code review from OpenAI Codex

Codex

brainstorm_codex

Architectural exploration, trade-offs, and system design ideation with Codex

Codex

plan_codex

Step-by-step implementation planning and checklist generation

🧪 Testing

# Run all Node.js MCP server & resilience tests
npm test

# Run all Go CLI detector & config tests
go test -v ./cli/...

📄 License & Acknowledgements

Created and architected by Marcos (H0wZy) under the MIT License.

This project unifies and enhances foundational work from the open-source MCP community:

This server cannot be deployed

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Maintenance

ActivityMaintained
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers

  • Boring Gemini

    Boring206
    F
    license
    Not graded
    quality
    C
    maintenance
    An autonomous AI development agent that enables full-stack coding, automated verification, RAG-powered code search, and quality assurance through MCP tools. Supports Gemini CLI, Claude Code CLI, with features like parallel verification, security scanning, and spec-driven development.
    5
    -

  • Multi-MCP

    religa
    A
    license
    A
    quality
    B
    maintenance
    A multi-model AI orchestration MCP server for automated code review and LLM-powered analysis, integrating with Claude Code and OpenCode to orchestrate multiple AI models for code quality checks, security analysis, and multi-agent consensus.
    6
    35
    MIT

  • all-agents-mcp

    Dokkabei97
    A
    license
    B
    quality
    F
    maintenance
    Enables orchestrating multiple AI CLI agents (Claude Code, Codex, Gemini CLI, Copilot CLI) through a unified MCP interface for task delegation, cross-agent comparison, and specialized tools like code review and debugging.
    14
    4 npm
    14
    MIT