OpenClaw Hackathon, Thu Feb 12th at MIT

For Serious Coders only You will need a Govt ID that matches your LuMa registration Please 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 controls Security + trust: sandboxing, tool permissions, secret handling, policy enforcement, audit logs, prompt/tool injection defenses Agent “app” primitives: games, marketplaces, social experiences, shared environments, agent-to-agent coordination, directories/registries DevEx + 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.