Moonshot Playground

Welcome to the Moonshot Playground

Welcome to the Moonshot Playground! This environment provides you with a fully running instance of Project Moonshot — an open-source LLM evaluation and red-teaming platform by the AI Verify Foundation (Singapore IMDA). It's designed for hands-on experimentation and learning in a safe, isolated environment.

Getting Started

With Moonshot, you can:

• Benchmark LLMs using standardized recipes and cookbooks.
• Red-team LLMs with automated attack modules to find safety gaps.
• Compare models side by side with consistent evaluation criteria.
• Chat interactively with configured LLM endpoints.

KodeKeys

To utilize Moonshot, you will require API Keys. You may obtain your own keys. To assist you, KodeKloud is providing AI Keys via KodeKeys. Please follow these instructions to retrieve the keys:

‍

1. Navigate to https://kodekloud.com/ai-playgrounds/kodekey
2. Click on Launch now, then select Start Playground.

You will need to:

• Select your model: Choose the AI Model required for your tasks.
• Base URL: Utilize this Base URL whenever you are using the key.
• API Key: This is the AI Key you will employ in your tasks.

Extending Sandbox Duration

To extend the duration of this sandbox environment, simply click on the (Extend lab by 15m) button located next to the timer.

When you're ready to conclude your session, click on the "End Sandbox" button in the Task Tab or the 🟥 button in the top right corner (visible after clicking on the toggle button) to gracefully terminate this sandbox environment.

‍Happy Evaluating!