You have three hackathon challenges to choose from:

• Build a purposeful AI agent: Build an AI-powered agent that solves a real-world problem by integrating with third-party platforms using Descope Outbound Apps.
• Build a secure MCP server: Transform your existing web, mobile, or API-based application into an AI-ready, agent-accessible MCP server using the Cequence AI Gateway.
• Build multi-agent systems: Build a secure, multi-agent system where autonomous agents communicate and collaborate across trust boundaries using Descope Inbound Apps.

For each Theme there will be a First, Second and Third-place prize.

• First Prize: $4000 + $75K in Daytona credits + 6 months of Vercel Pro
• Second Prize: $3000 + $50K in Daytona credits + 6 months of Vercel Pro
• Third Prize: $1500 + $10K in Daytona credits + 6 months of Vercel Pro

Please note that Daytona credits are only redeemable / usable by a single person per challenge winner. If a winning team comprises more than one person, we invite you to select who within your team gets the Daytona credits.

The hackathon submissions will be judged on set criteria by a panel of judges with extensive experience in agentic AI and MCP. The judging criteria are:

Utility & Relevance: Does your solution meaningfully support agent-based workflows?

• Addresses real-world needs or enhances agent functionality
• Applies to diverse scenarios such as formatting, tutoring, or language tasks
• Adds value to the broader ecosystem of autonomous agents

Creativity & Originality: Is your solution innovative and designed with agent interaction in mind?

• Introduces novel ideas or reimagines existing solutions
• Enhances agent autonomy, usability, or adaptability
• Offers advanced capabilities beyond basic functionality

Security & Access Control: Is your solution well-protected and clearly scoped for secure access?

• Implements secure authentication and authorization practices
• Properly restricts access using scoped permissions
• Manages which agents or services can connect and use the tools, enforcing trust and security boundaries

Technical Execution: Is your system technically sound and well-integrated?

• Utilizes products required for specific hackathon challenges
• Stable, complete, and reliably functional
• Effectively uses external integrations such as APIs, services, or agent clients
• Demonstrates a thoughtful architecture with clearly defined components

Connectivity & Compatibility: Can your server reliably communicate with external agent clients?

• Supports consistent and successful interactions with external agent environments
• Handles requests with low latency and minimal errors
• Responds correctly to common request types and protocols

Demonstration & Clarity: Does your submission clearly showcase how the system works in practice?

• Provides a clear, complete demonstration of the end-to-end workflow
• Highlights security enforcement and system interaction
• Easy to follow, visually and technically