NabehApp Documentation for AI Agents

NabehApp equips autonomous AI agents with dedicated, sandboxed email inboxes—allowing them to receive incoming mail and extract OTP verification codes in real-time via the Model Context Protocol (MCP) or modern REST APIs.

60-Second Quickstart #

Follow these 3 simple steps to connect your agent:

Connect with Cursor & Cline (Remote MCP) #

NabehApp officially supports the Remote Model Context Protocol (MCP). Add this to .cursor/mcp.json or Cline in VS Code:

{
  "mcpServers": {
    "nabeh-mail": {
      "type": "remote",
      "url": "https://portal.nabehapp.com/api/mcp",
      "headers": {
        "Authorization": "Bearer pn_agt_YOUR_AGENT_KEY",
        "X-Mail-Token": "pn_agt_YOUR_AGENT_KEY",
        "X-Mailbox": "your_agent@nabehapp.com"
      }
    }
  }
}

Connect with Claude Desktop #

Add to your claude_desktop_config.json file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json
• Linux: ~/.config/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "nabeh-agent-inbox": {
      "type": "remote",
      "url": "https://portal.nabehapp.com/api/mcp",
      "headers": {
        "Authorization": "Bearer pn_agt_YOUR_AGENT_KEY",
        "X-Mailbox": "your_agent@nabehapp.com"
      }
    }
  }
}

Environment Variables (.env) #

If you are building custom agents with Python, Node.js, LangChain, or CrewAI, add these variables to your .env file:

# NabehApp Agent Mail Configuration
NABEH_AGENT_KEY="pn_agt_YOUR_AGENT_KEY"
NABEH_MAILBOX="your_agent@nabehapp.com"
NABEH_MCP_URL="https://portal.nabehapp.com/api/mcp"

Direct Python API Call #

Simple Python snippet to fetch the latest OTP code directly via MCP tool invocation:

import os
import requests

AGENT_KEY = os.getenv("NABEH_AGENT_KEY", "pn_agt_YOUR_AGENT_KEY")
MAILBOX = os.getenv("NABEH_MAILBOX", "your_agent@nabehapp.com")

def get_latest_otp():
    """Extract the latest OTP verification code received by the agent"""
    url = "https://portal.nabehapp.com/api/mcp"
    payload = {
        "jsonrpc": "2.0",
        "id": "1",
        "method": "tools/call",
        "params": {
            "name": "get_latest_otp",
            "arguments": {
                "mailbox": MAILBOX
            }
        }
    }
    headers = {
        "Authorization": f"Bearer {AGENT_KEY}",
        "Content-Type": "application/json"
    }
    
    response = requests.post(url, json=payload, headers=headers)
    data = response.json()
    return data.get("result", {})

if __name__ == "__main__":
    result = get_latest_otp()
    print("Latest OTP Result:", result)

The 7 MCP Mail Tools Reference #

The NabehApp Remote MCP Server provides 7 specialized tools for full autonomous email capability:

1. get_latest_otp ⭐ (Primary Tool)

Instant sub-second extraction of the most recent OTP verification code with optional immediate consumption (consume: true).

2. list_incoming_emails

Lists incoming messages for the mailbox with message IDs, timestamps, sender details, and content snippets.

3. read_email

Retrieves the full MIME body, attachments, and decoded HTML/text content for a specific numeric email id.

4. send_email

Sends outbound or transactional email using verified domains and automatic idempotency key handling.

5. search_emails

Searches mailbox messages by keyword across subject lines, senders, and message bodies.

6. delete_email

Permanently deletes a processed message from the mailbox for privacy and hygiene.

7. clear_temporary_otps

Purges all temporary verification codes and ephemeral OTP records for the mailbox with a single command.

Security, Privacy & Isolation #

One-Time Key Reveal & SHA-256 Encryption

Following industry standards (similar to OpenAI and Anthropic), API keys are revealed to users exactly once upon creation. Once copied, plaintext keys are wiped from memory and permanently stored encrypted in D1 database records.

Agent Mailbox Sandboxing

Each agent mailbox operates in complete tenant isolation. Agents cannot inspect messages belonging to other mailboxes without explicit authorization, ensuring total safety when utilizing third-party autonomous AI swarms.

Frequently Asked Questions (FAQ) #