Agentic AI Human Approve Or Reject Agent — AI skill for Claude Code

Human-in-the-Loop Agent

A Streamlit application that demonstrates human approval workflows for sensitive AI actions using LangGraph's interrupt/resume mechanism.

Features

• Human Approval System: AI pauses before executing sensitive actions (send email, delete file, make purchase)
• Interrupt/Resume: Uses LangGraph's interrupt() to freeze execution and wait for user decision
• Conversation Memory: Maintains chat history with checkpoint persistence
• Multi-turn Conversations: Supports complex multi-step interactions with built-in context

Project Structure

.
├── app.py                 # Main Streamlit application
├── config.py             # Configuration & environment variables
├── tools.py              # Tool definitions (send_email, delete_file, make_purchase)
├── prompts.py            # System prompts for Claude
├── nodes.py              # Graph nodes (agent, approval, routing logic)
├── graph.py              # LangGraph state machine builder
├── llm.py                # LLM initialization with tools
├── models.py             # Type definitions
├── utils.py              # Utility functions
├── requirements.txt      # Python dependencies
├── .env.example          # Example environment variables
└── README.md             # This file

Setup

1. Install Dependencies

pip install -r requirements.txt

2. Configure Environment Variables

Create a `.env` file by copying `.env.example`:

cp .env.example .env

Then update with your actual credentials:

SMTP_USERNAME=your-email@gmail.com
SMTP_PASSWORD=your-app-password
SMTP_HOST=smtp.gmail.com
SMTP_PORT=465
SMTP_FROM=your-email@gmail.com

**Note**: For Gmail, use an [App Password](https://support.google.com/accounts/answer/185833), not your regular password.

3. Run the Application

streamlit run app.py

The app will open in your browser at `http://localhost:8501`.

How It Works

1. User Input: Type a message requesting an