Stop Fighting Your .env Files When Working with AI

Your API keys are locked away in .env files that your AI assistant can't see. When Claude needs to make an API call for you, you're stuck copy-pasting keys manually or watching it fail because it can't access your credentials. ServeMyAPI fixes this by storing your keys in macOS Keychain and making them available through natural language requests.

The Hidden Context Problem

Every macOS developer knows this frustration: .env files live in .gitignore for security, creating "hidden context" that breaks AI workflows. When you ask Claude to "deploy this to production" or "check the GitHub API for issues," it hits a wall because your keys are invisible.

Traditional approaches force you to choose between security and AI accessibility. ServeMyAPI gives you both.

What ServeMyAPI Actually Does

Instead of juggling multiple .env files across projects, you store API keys once in macOS Keychain and access them anywhere. Your AI assistant can request keys through natural language: "get my OpenAI API key" or "store this GitHub token as github_prod."

The keys stay secure in Keychain (requiring system authentication), but they're accessible to your AI tools when needed. No more manual key shuffling or broken workflows.

Key Benefits

Centralized Key Management: Store keys once, use everywhere. Update your OpenAI key in one place instead of hunting through 15 different projects.

Natural Language Access: Tell Claude "use my Stripe test key for this payment integration" and it retrieves the key directly through MCP.

Zero Plaintext Storage: Keys never touch config files or logs. Everything routes through macOS Keychain's secure storage.

CLI + MCP Integration: Manage keys from terminal (api-key store github_token ghp_xyz) or through any MCP-compatible client.

Real-World Use Cases

API Integration Development: Building a payment system? Store your Stripe test/live keys in Keychain. Claude can grab the right key based on context without you specifying which .env file to check.

Multi-Project Workflows: Working on three Next.js apps that all need the same Supabase key? Store it once, access it from any project through the MCP interface.

Deployment Automation: AI assistants can handle deployment tasks that require API keys without you manually providing credentials for each operation.

Code Reviews with Context: When Claude reviews your API integration code, it can understand the full context including which services you're connecting to, since keys are accessible rather than hidden.

Getting Started

ServeMyAPI works immediately with Claude Desktop and other MCP clients:

git clone https://github.com/Jktfe/serveMyAPI.git
cd serveMyAPI
npm install && npm run build

Add to your Claude Desktop config:

{
  "mcpServers": {
    "serveMyAPI": {
      "command": "node",
      "args": ["/path/to/servemyapi/dist/index.js"]
    }
  }
}

Store your first key via CLI:

api-key store openai_key sk-your-key-here

Now Claude can access it: "Using serveMyAPI, get my OpenAI key and make a completion request."

The server also runs standalone for HTTP access or integrates with Smithery for hosted deployment.

macOS-Only Trade-off

ServeMyAPI is macOS-specific because it relies on Keychain for security. While this limits cross-platform use, it provides genuine security that generic solutions can't match. For macOS developers working with AI assistants, the productivity gains outweigh platform limitations.

The roadmap includes investigating Windows/Linux credential storage, but the current macOS focus delivers a polished experience where it matters most.