MCP Enterprise Server (Model Context Protocol Gateway)

Author: Vivek Jaiswal (vivekjais16@gmail.com)
Role: Senior Software Engineer — Python | Django | FastAPI | Generative AI & Agentic AI

🏛️ Architectural Overview

MCP Enterprise Server is a high-performance, security-hardened implementation of Anthropic's Model Context Protocol (MCP) specification (JSON-RPC 2.0). It provides stateful AI agents, LLM tool-calling engines, and client interfaces (Claude Desktop, LangGraph supervisors, Cursor) with real-time access to production tools, enterprise databases, knowledge bases, and system observability metrics.

+-------------------------------------------------------------+
|               LLM Client / Autonomous Agent                 |
|             (Claude Desktop / LangGraph Router)             |
+------------------------------+------------------------------+
                               |
              JSON-RPC 2.0     |   SSE Stream / Stdio
                               v
+-------------------------------------------------------------+
|              MCP Enterprise Server (FastAPI)                |
|  +---------------------+  +-------------------------------+ |
|  | Bearer Auth & RBAC  |  | Rate Limiter (120 req/min)    | |
|  +---------------------+  +-------------------------------+ |
|  +--------------------------------------------------------+ |
|  |                JSON-RPC 2.0 Engine                     | |
|  |     (initialize, tools/*, resources/*, prompts/*)      | |
|  +--------------------------------------------------------+ |
|        |                  |                 |               |
|        v                  v                 v               |
|  +---------------+  +------------+  +-------------------+   |
|  | Database SQL  |  | Knowledge  |  | Sandboxed Python  |   |
|  | Safety Guard  |  | Search RRF |  | AST Inspector     |   |
|  +---------------+  +------------+  +-------------------+   |
+-------------------------------------------------------------+

Related MCP server: SentinelAI MCP Server

⚡ Core Capabilities

1. Enterprise Tools

• query_database: Parameterized SQL query execution with AST validation that blocks mutations (DROP, DELETE, UPDATE, ALTER) in read-only agent contexts.

• search_knowledge_base: Hybrid semantic retrieval across indexed technical documentation with vector scoring.

• system_metrics: Live host CPU, RAM, disk, and process telemetry inspection for automated SRE agents.

• execute_code_sandbox: Isolated Python execution with AST import inspection, blocking unsafe primitives (os, sys, eval, __import__).

2. Contextual Resources

• system://status — Live operational health and worker telemetry.

• database://schema — Relational schema reflection and column definitions.

• architecture://overview — Enterprise microservices topology.

3. Prompt Templates

• audit_sql_query — Pre-built prompt structure for SQL index optimization and injection audits.

• incident_investigation — Automated incident response runbook generation.

4. Transports

• Server-Sent Events (SSE) (/sse): High-throughput streaming transport for distributed microservices and web agents.

• Stdio Transport (src/transports/stdio.py): Standard I/O interface compatible with Claude Desktop and command-line LLM runners.

🚀 Quick Start

1. Clone & Setup

git clone https://github.com/vivekjais16/mcp-enterprise-server.git
cd mcp-enterprise-server

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

2. Run Test Suite

pytest tests/ -v

3. Launch Server

uvicorn src.main:app --host 0.0.0.0 --port 8080 --reload

🔒 Security & Authorization

All requests to direct JSON-RPC (/mcp/rpc) and SSE (/sse) require Bearer token or API key authentication:

curl -X POST http://localhost:8080/mcp/rpc \
  -H "Authorization: Bearer mcp_live_secret_key_prod_8920171244" \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/list"
  }'

🐳 Docker Deployment

docker-compose up -d --build