@jetkey/mcp-bridge

MCP bridge that multiplexes multiple Jetkey cloud servers into a single stdio interface with 7 universal tools.

Overview

The Jetkey MCP Bridge connects AI assistants to cloud infrastructure through the Model Context Protocol (MCP). It provides a unified interface to manage AWS, GCP, Azure, and other cloud services across multiple accounts—all through natural language.

Key Benefits:

• Unified Access: One bridge, multiple cloud accounts
• Constant Tool Count: 7 tools regardless of how many servers you have (50 servers = still 7 tools)
• Natural Language Search: Find cloud operations using semantic search
• Safety Controls: Dry-run previews and approval workflows for high-risk operations
• Auto-Updates: Background update checks with zero intervention

Installation

npm/npx

npx @jetkey/mcp-bridge YOUR_API_KEY

MCP Client Configuration

Add to your MCP client's configuration file:

{
  "mcpServers": {
    "jetkey": {
      "command": "npx",
      "args": ["-y", "@jetkey/mcp-bridge@latest", "YOUR_API_KEY"]
    }
  }
}

Common config file locations:

• Cursor: ~/.cursor/mcp.json
• Claude Desktop: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows)
• Other MCP clients: Refer to client documentation

Get your API key from: https://jetkey.com/user/api-keys

The 7 Universal Tools

1. servers

List all cloud MCP servers you have access to.

servers()
// Returns: { servers: [{ id, name, slug, bundleName }], count }

2. services

List cloud services available in a specific server.

services({ server_id: "df1e03e2-..." })
// Returns: { services: { aws: ["ec2", "s3", "lambda", ...] }, bundleName }

3. operations

List all operations for a service with pagination and filtering.

operations({
  server_id: "df1e03e2-...",
  service: "s3",
  limit: 50,        // optional, default 50, max 200
  offset: 0,        // optional, for pagination
  tags: ["list"],   // optional, filter by operation type
  search: "bucket"  // optional, filter by name/description
})
// Returns: operation details including required parameters

4. search

Search for tools using natural language queries.

search({
  query: "list S3 buckets",
  server_id: "df1e03e2-...",  // Required
  limit: 10,                   // optional
  threshold: 0.3               // optional, similarity threshold
})
// Returns: { tools: [{ toolId, name, description, similarity }] }

5. inspect

Get detailed parameter schemas and documentation for a tool.

inspect({
  server_id: "df1e03e2-...",
  tool_id: "aws.s3.list-buckets",
  mode: "summary"  // or "full"
})
// Returns: { requiredParams, commonParams, examples, hint }

6. execute

Execute a cloud operation.

execute({
  server_id: "df1e03e2-...",
  tool_id: "aws.s3.list-buckets",
  parameters: {},
  dryRun: true  // optional: preview without making changes
})
// Returns: operation result or dry-run preview

7. approval

Manage approval workflows for high-risk operations.

// List pending approvals
approval({ action: "list" })

// Check specific approval status
approval({ action: "status", approval_id: "abc123" })

// Execute an approved operation
approval({ action: "retry", approval_id: "abc123" })

Typical Workflow

1. "What cloud accounts do I have?"
   → servers()

2. "What services can I access in AWS Production?"
   → services({ server_id: "df1e03e2-..." })

3. "What S3 operations are available?"
   → operations({ server_id: "df1e03e2-...", service: "s3" })

4. "Find tools for listing buckets"
   → search({ query: "list S3 buckets", server_id: "df1e03e2-..." })

5. "What parameters does ListBuckets need?"
   → inspect({ server_id: "df1e03e2-...", tool_id: "aws.s3.list-buckets" })

6. "List my buckets"
   → execute({ server_id: "df1e03e2-...", tool_id: "aws.s3.list-buckets" })

Environment Variables

| Variable | Description | Default | |----------|-------------|---------| | JETKEY_MCP_URL | Base URL for Jetkey API | https://jetkey.com |

Example for local development:

{
  "mcpServers": {
    "jetkey": {
      "command": "npx",
      "args": ["-y", "@jetkey/mcp-bridge", "YOUR_API_KEY"],
      "env": {
        "JETKEY_MCP_URL": "http://localhost:3001"
      }
    }
  }
}

Auto-Update Mechanism

The bridge automatically checks for updates on startup:

• Checks NPM registry (throttled to once per 24 hours)
• Downloads updates in the background (non-blocking)
• Updates apply on next bridge restart
• Gracefully handles network failures

Check your version:

npx @jetkey/mcp-bridge --version

Architecture

MCP Client (Cursor, Claude Desktop, etc.)
    ↓ stdio
Jetkey MCP Bridge (7 universal tools)
    ↓ HTTP/JSON-RPC
Discovery API → [Server A, Server B, Server C, ...]
    ↓
Cloud Operations (AWS, GCP, Azure, etc.)

The bridge:

1. Discovers available servers via the Jetkey API
2. Exposes 7 universal tools that route to appropriate backends
3. Handles authentication, logging, and error handling

Scalability

| Servers | Tools | |---------|-------| | 2 | 7 | | 10 | 7 | | 50 | 7 | | 1000 | 7 |

Unlike per-server tool prefixing (which creates 4N tools for N servers), universal tools keep the interface constant.

Development

# Install dependencies
npm install

# Build
npm run build

# Run locally
node dist/index.js YOUR_API_KEY

Requirements

• Node.js >= 18.0.0
• Jetkey API key (from https://jetkey.com/user/api-keys)

License

MIT

Questions? Visit jetkey.com or email [email protected]