Skip to main content

Introducing the MongoDB-RAG Playground

· 3 min read
Michael Lynn
Michael Lynn
Developer Advocate @ MongoDB
Account creation form

🚀 What is the MongoDB-RAG Playground?​

The MongoDB-RAG Playground is an interactive UI built directly into the mongodb-rag library. Designed to provide developers with a quick and easy way to explore vector search, the Playground helps users visualize their documents, manage indexes, and run search queries—all without writing extra code.

Once a developer has set up their MongoDB Atlas cluster and ingested data, they can launch the Playground with a single command:

npx mongodb-rag playground

This command spins up a UI where developers can view documents, manage vector indexes, and perform vector searches, making it a powerful tool for experimentation and debugging.

🔧 Getting Started with the Playground​

To take full advantage of the Playground, follow these steps:

1️⃣ Initialize MongoDB-RAG​

First, install the mongodb-rag package if you haven’t already:

npm install mongodb-rag dotenv

Then, initialize a configuration file:

npx mongodb-rag init

This will generate a .mongodb-rag.json file where you can specify your MongoDB connection string, database name, and collection name.

2️⃣ Ingest Data​

Before running vector search, you need to ingest documents. This can be done using:

npx mongodb-rag ingest --file data.json

or by uploading a document directly from the Playground UI.

3️⃣ Create a Vector Search Index​

Once the data is loaded, create an Atlas Vector Search Index using:

npx mongodb-rag create-index

Alternatively, you can manage indexes from within the Playground UI.

4️⃣ Launch the Playground​

With the dataset and vector index in place, start the Playground:

npx mongodb-rag playground

This command launches a local UI, allowing you to explore your data in real time.

🖥️ Features of the Playground​

📂 Document Viewer​

  • Browse your stored documents with ease.
  • Quickly inspect metadata, embeddings, and content.

🔍 Search Panel​

  • Run vector search queries directly from the UI.
  • Experiment with different similarity metrics (cosine, dot-product, Euclidean).

⚙️ Index Management​

  • View existing indexes.
  • Create new vector search indexes.
  • Configure embedding models and dimensions.

📡 Live Updates​

  • Real-time indexing and search updates via WebSockets.
  • Instant feedback on configuration changes.

🔥 Why Use the Playground?​

  • No need to build a frontend – The Playground provides an out-of-the-box UI for MongoDB vector search.
  • Debugging made easy – Quickly inspect documents and indexes without writing queries manually.
  • Seamless experimentation – Tweak search parameters and embeddings dynamically.

🎯 Try It Today​

The MongoDB-RAG Playground is the fastest way to interact with MongoDB Atlas Vector Search. Whether you're building an AI-powered search engine or testing RAG (Retrieval-Augmented Generation) workflows, the Playground gives you full control over your data and indexes in a simple UI.

Ready to get started?

Run:

npx mongodb-rag playground

And start exploring the power of MongoDB Vector Search today! 🚀