In the video, the creator showcases the Gemini 2.5 model and Cloud Code by setting up an MCP server that enables collaboration between Claude 3.7 and Gemini to develop a QWOP-style limb runner game. Despite encountering challenges with limb controls and balance, the creator successfully integrates suggestions from Gemini to enhance the game’s mechanics and plans to share the project on GitHub for community engagement.
In the video, the creator expresses their enthusiasm for the Gemini 2.5 model, which they consider the best large language model available. They also share their fondness for Cloud Code and its agentic workflow. To explore the potential of combining these two technologies, the creator sets up an MCP server that connects to the Gemini 2.5 Pro API, allowing Claude 3.7 to collaborate with Gemini. The goal is to create a simple game inspired by the QWOP limb control game, where players manage a character’s limbs to navigate challenges.
The creator begins by outlining their plan to develop a QWOP-style limb runner game using 3JS, focusing on a primarily 2D experience with some 3D elements. They prompt Claude to ask Gemini for a detailed implementation plan for the game. After sending the request, they receive a comprehensive response from Gemini, which includes a game concept, technological stack, and development phases. The creator expresses excitement about the plan and decides to proceed with coding the game immediately, rather than waiting for an extended development timeline.
As the development progresses, the creator runs into issues with the game, particularly with the limb controls not functioning as intended. They decide to send relevant code snippets to Gemini for assistance in troubleshooting the problem. This collaboration allows Claude and Gemini to work together to identify and resolve the issues with the character’s limb movements, enhancing the overall development process.
After making adjustments based on Gemini’s suggestions, the creator tests the game again. They find that while the character has improved mobility, it still struggles with balance and stability. The creator experiments with adding arms to the character to see if it helps with balance. They share their experiences of tweaking parameters and adjusting the physics, noting that while the game is still a work in progress, it has become more enjoyable to play.
In conclusion, the creator expresses satisfaction with the collaboration between Claude and Gemini through the MCP server, highlighting the potential for further exploration of this technology combination. They plan to share the game on GitHub for others to experiment with and improve upon. The video serves as both a demonstration of the collaborative capabilities of AI models and an invitation for viewers to engage with the project and contribute to its development.