Title: ChatGPT plugins & LangChain Tags: [chatgpt, langchain, plugins]

LangChain AI plugins

This directory contains an example of how to integrate public ChatGPT plugins with LangChain and Chainlit. It demonstrates the setup of a simple Chainlit application that utilizes an AI plugin to enhance the capabilities of a LangChain agent.

Description

The app.py script initializes a LangChain agent with a ChatGPT model and an external AI plugin. The AI plugin is loaded from a specified URL and combined with other tools to create a versatile agent. This agent can then process incoming messages and provide responses using both its built-in capabilities and the additional functionality provided by the AI plugin.

Quickstart

To get started with this example:

1. Ensure you have Chainlit and LangChain installed in your environment.
2. Clone this repository and navigate to the chainlit-cookbook/langchain-aiplugins directory.
3. Run the app.py script to start the Chainlit application.
4. Interact with the application through the Chainlit interface to see the LangChain agent and AI plugin in action.

Code Definitions

• AIPluginTool.from_plugin_url: Loads an AI plugin from the provided URL.
• ChatOpenAI: Initializes a ChatGPT model with specified parameters.
• load_tools: Loads additional tools that can be used by the agent.
• initialize_agent: Creates a LangChain agent with the loaded tools and AI plugin.
• AgentExecutor: Executes the agent's actions based on incoming messages.
• cl.on_chat_start: Decorator to initialize the agent when the chat starts.
• cl.on_message: Decorator to define the main message handling loop of the application.

Follow the instructions in the main readme to set up your environment and run the example.