Deepmind

📘 LangChain Agent with Tools, Memory & Tavily Search (LangChain v1)

A beginner-friendly AI Agent built using LangChain v1, OpenAI, Tavily Search, and custom tools. Supports:

✅ Tool Calling ✅ Short-Term Memory (Chat History) ✅ Web Search using Tavily ✅ Weather Tool ✅ Datetime Tool ✅ Interactive CLI Chat ✅ Clean & simple LangChain v1 setup

🚀 Features 🔧 Built-in Tools

get_current_datetime → Returns the current system time

get_weather(city) → Weather using the free wttr.in API (no key needed)

TavilySearchResults → Web search with Tavily

🧠 Memory System

Stores up to 10 recent message pairs (20 messages total)

Maintains context across conversations

Automatically formats memory for LangChain v1

🤖 Agent System

Uses ChatOpenAI (gpt-4o-mini)

Supports structured tool calling (LangChain v1 standard)

Handles parsing errors

Verbose mode enabled for debugging

📦 Requirements

Install using your Python 3.10 environment:

pip install langchain==0.3.14 pip install langchain-openai==0.2.14 pip install langchain-core==0.3.29 pip install langchain-community==0.3.14 pip install tavily-python==0.5.0 pip install python-dotenv==1.0.1 pip install requests==2.32.3 pip install openai>=1.0.0 pip install pydantic>=2.0.0

Or, add everything inside requirements.txt and run:

pip install -r requirements.txt

🔑 Environment Variables

Create a file called .env in the same folder as your script:

OPENAI_API_KEY=your-openai-key-here TAVILY_API_KEY=your-tavily-key-here

Both keys are required:

OPENAI_API_KEY → for the LLM

TAVILY_API_KEY → for web search

▶️ How to Run 1️⃣ Activate venv venv\Scripts\activate

2️⃣ Run the agent python main.py

You will see:

============================================================ LangChain Agent with Tools and Memory (v1)

📋 Loading API keys... 🤖 Initializing agent... ✅ Agent ready!

💬 Usage (Inside the Chat)

Enter messages normally:

You: what is the time now? 🤖 Agent: The current time is...

⏱ Useful Commands Command Action quit / exit / q Stop the agent history Show recent chat history clear Clear memory 🛠 Project Structure project/ │ ├── main.py # Your agent code ├── .env # API keys ├── requirements.txt # Dependencies └── README.md # Documentation

📙 How the Agent Works

1. Loads tools
2. Loads prompt with memory placeholders
3. Uses create_tool_calling_agent() (correct v1 method)
4. Uses AgentExecutor to run
5. Stores every human/assistant message in memory
6. Trims memory to last 10 interactions
7. Responds using LLM — with tool support when needed

Done !