The Gradio Agents & Model Context Protocol (MCP) Hackathon 2025 is an online, global, week-long event scheduled from June 2 to June 10, 2025. The hackathon's primary objective is to advance the capabilities of AI Agents and the Model Context Protocol (MCP) by encouraging participants to develop innovative Gradio-based applications. MCP is an open protocol designed to standardize how applications provide contextual information to Large Language Models (LLMs), thereby enabling LLMs to interact more effectively with the external world.

Participants are required to join the hackathon's organizing entity on Hugging Face, register through a provided form, and submit their projects as Gradio Spaces to this organization by the final submission deadline of June 10, 2025, 11:59 PM UTC. Submissions can be individual or by teams of 2-5 members.

The hackathon features three distinct tracks, each with specific technical requirements and submission guidelines:

1. Track 1: MCP Tool / Server
   • Objective: To develop a Gradio application that functions as an MCP Server or Tool. This involves extending the capabilities of LLMs by enabling them to interact with a Gradio app that serves a specific task.
   • Methodology: The core technical concept relies on the principle that a Gradio application can be configured to act as an MCP server, thereby providing structured context to LLMs. This involves implementing the necessary interfaces and functionalities within the Gradio app to adhere to the MCP specification.
   • Submission Requirements: The project must be a Gradio app Space, marked with the tag "mcp-server-track" in its README.md file. A video recording demonstrating the MCP server's operation with an MCP client (e.g., Claude Desktop, Cursor, Tiny Agents, or another Gradio app functioning as a client) must be included or linked in the README.md.

2. Track 2: Custom Components for Agents
   • Objective: To create novel Gradio custom components that enhance the functionality, intuitiveness, or power of AI agent interfaces.
   • Methodology: This track focuses on front-end and component development within the Gradio ecosystem. Participants are expected to leverage Python and JavaScript skills to build custom UI elements for Gradio 5. The development process typically involves defining the component's behavior and appearance, and then publishing it as a Gradio Space using the gradio cc publish command.
   • Submission Requirements: The submission must be a custom Gradio component published as a Space, tagged with "custom-component-track" in its README.md. A link to a Gradio app or a recording showcasing the custom component in action is mandatory in the README.md.

3. Track 3: Agentic Demo Showcase
   • Objective: To showcase the power and creativity of AI agents through a comprehensive Gradio application, regardless of whether it directly utilizes MCP tools.
   • Methodology: This track encourages broad innovation in AI agent systems. Technical implementation can vary widely, from sophisticated multi-agent architectures to simpler agentic workflows, all encapsulated within a Gradio app for demonstration. The focus is on the practical application and impact of AI agents.
   • Submission Requirements: The project must be a complete Gradio application published as a Space, tagged with "agent-demo-track" in its README.md. A video overview explaining the application's usage and purpose must be included or linked in the README.md.

The hackathon offers a significant prize pool of over $16,500 in cash, with$2,500 for first place and $500 for second place in each track. Additionally, special awards are available, including a$5,000 Modal Labs Choice Award, a $2,000 Mistral AI Choice Award in API credits, and a$1,000 LlamaIndex Choice Award, among others. Participants also receive substantial free compute and API credits from sponsors, totaling over $1 Million, including$250 in GPU/CPU compute credits from Modal Labs and $25 in API credits from Hugging Face for every participant. Credits from other sponsors like Nebius, Anthropic, OpenAI, Hyperbolic Labs, MistralAI, and Sambanova.AI are provided to a limited number of early participants.

Projects will be judged based on innovation, technical implementation, usability, and overall impact, with community likes on Spaces also factoring into the evaluation. Support is provided through office hours with experts and a dedicated Discord channel. No prior experience with Agents or MCP is required for participation, as resources and support will be available.