WillSmith: A LangSmith Open-Source Alternative 🚀

WillSmith is an open-source alternative to LangSmith, designed to provide a more cost-effective and flexible solution for monitoring, debugging, and optimizing applications based on large language models (LLMs). It supports easy deployment via Docker and features a sleek user interface.

Features ✨

Chain Call Monitoring: Track and record every model call's input, output, errors, and performance metrics for easy optimization.
User-Friendly UI: With a clean interface built using Shadcn UI, quickly browse and analyze stored data.
MongoDB Data Storage: Efficiently store and query chain call and debugging data with MongoDB.
Docker Deployment: Easily get started using the provided Docker setup.

Getting Started ⚡

1. Start with Docker Compose 🐳

This project provides a docker-compose.yaml file that includes everything you need to run the app and MongoDB. Make sure you have Docker and Docker Compose installed.

First, set up your environment variables in a .env file:

# .env
MONGODB_URI=mongodb://your-mongo-uri:27017/willsmith

Then, simply run:

docker-compose up -d

Docker Compose will start both the WillSmith app and the MongoDB instance. The app will be available at http://localhost:3000.

2. Access the App 🌐

Once the app is running, open your browser and visit http://localhost:3000 to access the UI and start managing your application data.

LangChain Integration 🔗

To integrate WillSmith with your LangChain project for chain call monitoring, just add the following environment variables to your .env file:

# In your LangChain project's .env file
LANGCHAIN_ENDPOINT=http://localhost:3000  # The service URL after Docker starts
LANGCHAIN_TRACING_V2=True
LANGCHAIN_API_KEY=your_api_key  # Any value, no key validation yet
LANGCHAIN_PROJECT=project_name  # Your custom project name

With these settings, LangChain will automatically send chain call data to WillSmith, which will store it in MongoDB.

Development Roadmap 🛠️

We have an exciting set of features in our development pipeline. Here’s what’s coming next:

Project Querying 🗂️ — Completed
Root Record Querying 📝 — Completed
Tree-Structured Detailed Record Querying 🌲 — In progress
Enhanced Database Design 📊 — Coming soon (refactor data into multiple collections)
Auto-Update for Record Queries 🔄 — Coming soon (dynamic query updates for latest records)