MCP 服务器

OmniForge MCP

An enterprise-grade MCP server that enables AI coding assistants to securely connect with external tools, APIs, databases, and cloud services through a unified interface, offering structured engineering workflows and multi-client support.

README

<h1 align="center">OmniForge MCP</h1>

Structured Engineering Development MCP Server for AI Coding Assistants

📌 Product Overview

OmniForge MCP is an enterprise-grade Model Context Protocol (MCP) platform created by Ankur Gaurav that enables AI coding assistants and autonomous agents to securely connect with external tools, APIs, databases, file systems, cloud services, and business applications through a unified, standardized interface.

The platform standardizes the way developer teams plan, build, test, and ship software by offering a structured, reliable, and repeatable engineering workflow.

🚀 Why OmniForge MCP?

Modern AI applications require reliable context and real-time execution capabilities. OmniForge MCP bridges the gap between Large Language Models (LLMs) and enterprise infrastructure by offering:

Structured Workflow: Follow a proven software engineering process from start to finish.
Engineered for Developers: Built to support real-world development practices and modern IDEs.
Consistent & Reliable: Ensure consistency across projects, repositories, and developer teams.
Multi-Client Installation: Single command omniforge install --all configures Claude Desktop, Cursor IDE, VS Code, OpenAI Codex, ChatGPT, Gemini, and Antigravity.
Accelerate Delivery: Reduce noise, focus on building, and deliver with confidence.

✨ Features & Capabilities

Feature	Status	Description
STDIO Transport	✅ Production Ready	Standard I/O line protocol for desktop clients (Claude Desktop, Cursor, VSCode)
HTTP & SSE Transports	✅ Production Ready	High-concurrency REST endpoints (`/rpc`, `/health`) and streaming event feeds (`/sse`)
Tool Registry	✅ Production Ready	Exposes 14 agent skills and 10 engineering workflows as invokable MCP Tools
Prompt Engine	✅ Production Ready	10 enterprise prompt library templates with variable interpolation (`{{var}}`)
Resource Registry	✅ Production Ready	Standardized `omniforge://` URI schemas for architecture docs and catalogs
CLI Control Suite	✅ Production Ready	12 operational subcommands (`init`, `doctor`, `install`, `search`, `workflow`, etc.)
Multi-Client Installers	✅ Production Ready	Automated setup for Claude, Cursor, VS Code, Codex, ChatGPT, Gemini, and Antigravity
Plugin Ecosystem	✅ Production Ready	Dynamic plugin manager with `git-enhancer` and `security-auditor`
Authentication Middleware	✅ Production Ready	API Key and Bearer token security hooks for HTTP & SSE endpoints
TypeScript & Python SDKs	✅ Production Ready	Strongly-typed client libraries for TypeScript and Python

🏗️ Architecture Overview

OmniForge MCP follows Hexagonal Architecture principles, separating core business logic from external transports and integrations.

graph TD
    subgraph Clients["AI Clients & IDEs"]
        Claude["Claude Desktop"]
        Cursor["Cursor IDE"]
        VSCode["VSCode Extension"]
        Codex["OpenAI Codex"]
        ChatGPT["ChatGPT Actions"]
        Gemini["Google Gemini"]
        Antigravity["Google Antigravity"]
    end

    subgraph Transports["Multi-Transport Gateway"]
        STDIO["STDIO Transport"]
        HTTP["HTTP REST (/rpc, /health)"]
        SSE["SSE Event Stream (/sse)"]
    end

    subgraph CoreEngine["OmniForge Core Engine"]
        Auth["Auth & Security Middleware"]
        ToolReg["Tool Registry (14 Skills + 10 Workflows)"]
        PromptReg["Prompt Registry (10 Templates)"]
        ResourceReg["Resource Registry (omniforge://)"]
        Runner["Workflow Execution Engine"]
        PluginMgr["Plugin Manager & Marketplace"]
    end

    Clients -->|STDIO / HTTP / SSE| Transports
    Transports --> Auth
    Auth --> ToolReg
    Auth --> PromptReg
    Auth --> ResourceReg
    ToolReg --> Runner
    ToolReg --> PluginMgr

💻 Supported Clients

Claude Desktop (Anthropic MCP Client)
Cursor IDE
VS Code Extensions
OpenAI Codex CLI
ChatGPT Actions & Custom GPTs
Google Gemini CLI
Google Antigravity Agent
LangChain & LlamaIndex Agents
Custom Autonomous AI Agents

📦 Installation

Install the OmniForge CLI globally via npm:

npm install -g omniforge

Or run via npx:

npx omniforge doctor

Install MCP integration into all supported AI clients:

omniforge install --all

⚡ Quick Start

1. System Diagnostic Health Check

Verify system environment, Node.js version, Git status, write permissions, and MCP server readiness:

omniforge doctor

2. Initialize Workspace

Generate .omniforge/config.json in your project directory:

omniforge init

3. Install AI Client Configurations

Auto-detect and generate MCP configuration files:

omniforge install --all

4. Start Production MCP Server

Start the server in HTTP/SSE mode (default port 8080):

npm run mcp:start

For STDIO mode (IDE integration):

node mcp/dist/server.js --stdio

📂 Folder Structure

omniforge/
├── apps/               # Modern Documentation & Interactive Dashboard App
├── packages/           # Monorepo Core & Client SDKs
│   ├── core/           # ConfigManager, Logger, Telemetry, PluginLoader, UpdateManager
│   └── sdk/            # TypeScript & Python Client SDKs
├── mcp/                # Production MCP Server (STDIO, HTTP, SSE)
│   ├── mcp.json        # MCP Manifest Descriptor
│   └── src/            # Transports, Handlers, Middleware
├── cli/                # OmniForge CLI Control Suite & Installers
├── skills/             # 14 Standardized & Categorized Agent Skills
├── prompts/            # 10 Enterprise Prompt Library Templates
├── workflows/          # 10 Executable Engineering Workflows
├── templates/          # Generator Templates (Project, Skill, Prompt, Plugin)
├── plugins/            # Dynamic Plugin System & Marketplace Architecture
├── docs/               # 16 Complete Documentation Guides & Tutorials
├── examples/           # Executable Demos & SDK Usage
├── tests/              # Vitest Test Suite (18 Test Files, 52 Tests)
├── sprints/            # 11 Sprint Deliverables Tracking (Sprint 1 to 11)
└── docker/             # Dockerfile & Docker Compose Configs

🗺️ Roadmap

All 11 development sprints have been fully completed:

Sprint	Subfolder	Key Deliverables	Status
Sprint 1	`sprints/sprint-1/`	Monorepo Setup & Core Layout	✅ Completed
Sprint 2	`sprints/sprint-2/`	Architecture & Error Hierarchy	✅ Completed
Sprint 3	`sprints/sprint-3/`	Multi-Transport MCP Server (STDIO, HTTP, SSE)	✅ Completed
Sprint 4	`sprints/sprint-4/`	14 Agent Skills Migration & Validation	✅ Completed
Sprint 5	`sprints/sprint-5/`	Enterprise Prompt Library & Prompt Engine	✅ Completed
Sprint 6	`sprints/sprint-6/`	OmniForge CLI Suite & Master Rebranding	✅ Completed
Sprint 7	`sprints/sprint-7/`	Plugin Loader Ecosystem & Marketplace Schema	✅ Completed
Sprint 8	`sprints/sprint-8/`	Multi-Client MCP Installer Suite (7 AI Clients)	✅ Completed
Sprint 9	`sprints/sprint-9/`	Complete 16 Documentation Guides & Portal	✅ Completed
Sprint 10	`sprints/sprint-10/`	Performance, Regression & Coverage Test Suite	✅ Completed
Sprint 11	`sprints/sprint-11/`	v1.0.0 Production Release & Public Launch	✅ Completed

🤝 Contributing

Contributions are welcome! Please read CONTRIBUTING.md and CODE_OF_CONDUCT.md before submitting pull requests.

git clone https://github.com/ankur-gaurav161418/omniforge.git
npm install
npm run build
npm test

📄 License

Distributed under the MIT License.
Copyright (c) 2026 Ankur Gaurav (https://github.com/ankur-gaurav161418).

💬 Support & Community

Repository: https://github.com/ankur-gaurav161418/omniforge
Issues: Submit a Bug or Feature Request
Security Policy: SECURITY.md
Author: Ankur Gaurav (https://github.com/ankur-gaurav161418)