Narad Email Agent

🪄 Narad — Your AI-Powered Job Search Engine

Not just an email tool. A full AI-powered career management system that drafts, tracks, and manages your entire job search pipeline — available as a CLI or as a plug-and-play MCP Server inside your AI Assistant.

Narad is powered by Google Gemini 2.0 Flash. It reads your CV, matches your skills to a Job Description, drafts a hyper-personalized email, tracks every application in a local CRM, preps you for interviews, and can now be used directly from inside Antigravity, Cursor, or Claude Desktop — without ever opening a terminal.

✨ Features

🛠️ Tech Stack

• AI Model: Google Gemini 2.0 Flash (google-generativeai)
• Email: Gmail SMTP (Sending) + Gmail IMAP (Reading)
• PDF Reading: pypdf
• Local CRM Database: SQLite (built-in Python, no server needed)
• MCP Protocol: fastmcp (Model Context Protocol)
• Environment: python-dotenv
• Language: Python 3.x

⚡ Quick Setup

1️⃣ Clone & Create Virtual Environment

git clone https://github.com/karanshelar8775/Narad-Email-Agent.git
cd Narad-Email-Agent
python -m venv venv

# Activate (Windows)
.\venv\Scripts\Activate

# Install dependencies
pip install -r requirements.txt

2️⃣ Configure Your .env

cp .env.example .env

Open .env and fill in:

# ── Gmail Credentials ─────────────────────────────
EMAIL_ADDRESS="[email protected]"
EMAIL_PASSWORD="your_16_digit_app_password"
SMTP_SERVER="smtp.gmail.com"
SMTP_PORT=587
IMAP_SERVER="imap.gmail.com"

# ── Gemini API ─────────────────────────────────────
GEMINI_API_KEY="your_gemini_api_key_here"

# ── User Profile (For Personalized Drafting) ───────
USER_NAME="Your Full Name"
USER_UNIVERSITY="Your University"
USER_YEAR="Final Year"
USER_MAJOR="Your Major"
USER_PORTFOLIO="https://yourportfolio.vercel.app/"

💡 Gmail App Password: Google Account → Security → App Passwords

💡 Gemini API Key: Google AI Studio

3️⃣ Add Your Resume(s)

Drop your CV/Resume PDF(s) into the resumes/ folder. Narad will auto-detect all of them.

🎮 Mode 1: CLI Usage (Classic)

python main.py

============================================================
   🪄  NARAD | Your AI-Powered Job Search Engine
============================================================
Commands: 'send', 'check', 'summarize', 'stats', 'search', 'followup', 'interview', 'help', 'exit'

🔌 Mode 2: MCP Server (AI Assistant Integration)

This is the Power User mode. Narad can be connected to Antigravity, Cursor, or Claude Desktop and used directly from your AI chat — no terminal needed.

✅ Step 1: Verify mcp_config.json exists

The file should be at:

C:\Users\<YourName>\.gemini\antigravity\mcp_config.json

With this content:

{
    "mcpServers": {
        "narad": {
            "command": "C:\\Users\\<YourName>\\Desktop\\Narad-Email-Agent\\venv\\Scripts\\python.exe",
            "args": [
                "C:\\Users\\<YourName>\\Desktop\\Narad-Email-Agent\\mcp_server.py"
            ],
            "env": {
                "PYTHONPATH": "C:\\Users\\<YourName>\\Desktop\\Narad-Email-Agent"
            }
        }
    }
}

⚠️ Replace <YourName> with your actual Windows username.

✅ Step 2: Enable in Your AI Assistant

For Antigravity (VSCode):

1. Open Settings → Customizations
2. Find "INSTALLED MCP SERVERS"
3. Click Refresh 🔄 (close and reopen if needed)
4. narad should appear with a blue toggle ✅

For Cursor:

1. Open Cursor Settings → Features → MCP
2. Click "Add New MCP Server"
3. Name: narad, Command: the python path above

For Claude Desktop: Add the same block to claude_desktop_config.json in your Claude app data folder.

✅ Step 3: Use it with Natural Language!

Once connected, your AI Assistant can access all of Narad's tools. Just talk naturally:

🧰 MCP Tools Exposed

📁 Project Structure

Narad-Email-Agent/
├── agents/
│   ├── base_agent.py         # Base class for all agents
│   └── email_agent.py        # SMTP + IMAP + HTML formatter + bold rendering
├── core/
│   ├── gemini_client.py      # Gemini 2.0 Flash API wrapper
│   ├── composer.py           # AI drafting engine (CV + JD + Portfolio aware)
│   ├── database.py           # SQLite CRM for tracking applications
│   ├── summarizer.py         # AI inbox summarization
│   └── reader.py             # Inbox fetching and formatting
├── resumes/                  # Drop your resume PDFs here (Multi-CV support!)
│   └── YourResume.pdf
├── utils/
│   └── helpers.py            # Shared utility functions
├── main.py                   # CLI entry point (Mode 1)
├── mcp_server.py             # MCP Server entry point (Mode 2) ← NEW
├── narad_crm.db              # Auto-created local CRM database (gitignored)
├── requirements.txt          # Python dependencies
├── .env.example              # Safe credentials template
├── .gitignore                # Keeps credentials and DB out of GitHub
└── README.md                 # This file

🔐 Security

• .env is in .gitignore — your credentials are never pushed to GitHub.
• narad_crm.db is in .gitignore — your job search data stays private.
• Uses Gmail App Passwords, not your main Google password.
• All processing happens locally on your machine.

👨‍💻 Author

Karan Rohidas Shelar — Generative AI Developer

Built with ❤️ and Gemini 2.0 Flash — Because applying to jobs manually is so 2023.