Agent Identity Protocol (AIP)

# LangChain + Agent Identity Protocol Integration This example demonstrates the difference between a standard "Unsafe" tool and a "Secure" tool using Agent Identity. ## The Problem Standard LangChain tools execute immediately. If an LLM is hijacked or hallucinates, it can delete files or spend money instantly. ## The Solution We wrap the tool in an **Identity Layer**. The tool forces the agent to cryptographically sign the intent *before* execution. ## How to Run the Demo 1. **Install Dependencies:** ```bash pip install langchain pydantic ``` 2. **Run the Comparison:** ```bash python secure_tool.py ``` ## Expected Output You will see two scenarios: 1. **The Unsafe Way:** Money moves immediately with a warning. 2. **The Secure Way:** You will see the `🔐 AIP: Agent is signing action...` step before the bank accepts the transfer.