For Serious Coders onlyYou will need a Govt ID that matches your LuMa registrationPlease do not join if your registration is not approved
We’re hosting a hackathon for people who want to build the missing infrastructure layer for AI agents.
Agents are getting easier to spin up, but the hard parts are still unsolved: security, permissions, reliability, deployment, observability, safe tool use, and the apps/experiences that make agents actually useful. This hackathon is about shipping real building blocks—things other agents (and other builders) can plug into.
You can build for OpenClaw-style agents, or for any agent framework. The goal is interoperability: clean APIs, clear contracts, testable behavior, and sensible safety boundaries.
What you might build:
Agent infrastructure: deployment templates, agent hosting, state/memory patterns, job queues, scheduling, background tasks, cost controlsSecurity + trust: sandboxing, tool permissions, secret handling, policy enforcement, audit logs, prompt/tool injection defensesAgent “app” primitives: games, marketplaces, social experiences, shared environments, agent-to-agent coordination, directories/registriesDevEx + reliability: debugging tools, eval harnesses, traces, replay, failure recovery, monitoring dashboards
This is for students who like systems, product, and applied security and who want to ship something that could become part of the agentic web stack.
