Course Wiki

Configuration — AI Dev Tools Zoomcamp Module 5

## Codex ### CODEX + MCP Setup (VS Code — Linux)

AI Dev Tools Zoomcamp › Module 5: Coding Agent Building Blocks: Reusable Skills and Specialized Subagents › Configuration

Notes

Configuration

GitHub Copilot

…

Codex

CODEX + MCP Setup (VS Code — Linux)

These steps assume a Linux environment. Paths and commands may differ on Windows or macOS.

1. Install Codex Extension and Sign In

• Install OpenAI Codex from the VS Code Marketplace
• Sign in with your OpenAI account

2. Install the Codex CLI (Required)

Run in a terminal:

    npm install -g @openai/codex
    codex --version

Installing the CLI initializes the Codex config and config.toml.

3. Open Codex MCP Settings

• Open the Codex tab in VS Code
• Click the Settings (⚙️) button
• This opens the config.toml file used by Codex

4. Configure the MCP Server (Linux)

Add the following to config.toml:

    [codex.mcp_servers.hw3-mcp]
    command = "uv"
    args = [
      "--directory",
      "/path/to/ai-dev-tools-zoomcamp/05-agent-capabilities",
      "run",
      "python",
      "main.py"
    ]
    startup_timeout_sec = 10
    tool_timeout_sec = 60
    enabled = true

Note on server name
hw3-mcp is just a configuration label. It can be renamed to anything meaningful (e.g. local-mcp, mcp-tools). The name only needs to be unique and does not need to match the project folder or any code.

5. Trust the Project

Also add:

    [projects."/path/to/ai-dev-tools-zoomcamp"]
    trust_level = "trusted"

6. Expected Behavior

• MCP tools are available to Codex
• If the server is tools-only, Codex will report no MCP resources (this is expected)

Cursor

…

Antigravity

Prerequisites

• Antigravity AI Assistant (Google Deepmind)
• Node.js and npm installed (for most MCP servers)
• API keys for third-party MCP servers (if required)

Configuration File Location

Antigravity reads MCP server configurations from a JSON file. The configuration file should be named mcp.json and placed in your project directory.

Basic Configuration Format

Create an mcp.json file with the following structure:

{
    "mcpServers": {
        "server-name": {
            "command": "command-to-run",
            "args": [
                "argument1",
                "argument2"
            ]
        }
    }
}

Example: Adding Context7 MCP Server

Context7 provides access to library documentation and code search capabilities.

Step 1: Create mcp.json

{
    "mcpServers": {
        "context7": {
            "command": "npx",
            "args": [
                "-y",
                "@upstash/context7-mcp@latest"
            ]
        }
    }
}

Step 2: Set Up API Key (if required)

Some MCP servers like Context7 require authentication:

On Windows (PowerShell):

$env:CONTEXT7_API_KEY="your-api-key-here"

On Linux/Mac:

export CONTEXT7_API_KEY="your-api-key-here"

Permanent Setup (Windows): Add to your PowerShell profile or set as system environment variable.

Step 3: Restart Antigravity

After adding the MCP configuration, restart Antigravity to load the new MCP server.

Step 4: Verify MCP Server Connection

Once Antigravity restarts, you can verify that your MCP server is connected:

1. In Antigravity UI: Click the three dots (⋮) menu in the interface
2. Look for “MCP Servers” or similar option that shows connected servers
3. You should see your configured server(s) listed (e.g., “context7”)

[!IMPORTANT] If your MCP server doesn’t appear:

• Verify the mcp.json file is in the correct location
• Check for JSON syntax errors (use a JSON validator)
• Ensure Node.js and npm are installed and accessible from command line
• Check that the server package name is correct (e.g., @upstash/context7-mcp@latest)
• Review Antigravity’s logs for any error messages
• Try running the MCP command manually in your terminal to test if it works:

  npx -y @upstash/context7-mcp@latest
      

[!TIP] If Context7 or anoth

Key concepts

• OpenAI API
• MCP

Related notes

• aidt-m01-ai-native-developer-workflow
• aidt-m02-build-and-ship-an-ai-assisted-full-stack-app
• aidt-m03-test-containerize-and-deploy-an-ai-assisted-app
• aidt-m04-devops-and-observability-for-ai-built-apps
• aidt-m05-module-5-coding-agent-capabilities-mcp-skills-pl
• mlz-m01-model-selection-process

Sources

• Lesson file