Session - Lagos Walrus Memory x AI Prompt

The prize pool is $300 WAL. That's not really the point. The point is that by the end of this hackathon, you'll have actually built an AI agent that remembers — not read about it, not watched a demo, built one yourself, even with no knowledge of coding or web3. This is for everyone.Walrus Memory is what makes that possible, and this is a hands-on crash course in using it. But building it for yourself is only half of it — you'll also write down the prompt that made it work, and share it with us, so anyone else can pick it up and build the same thing. Sharing is caring: the best prompts here could end up helping people who weren't even in the room.Bring an idea, bring curiosity, and spend a day learning how AI agents work by building one that works for you — then hand it forward.This is a local event, so you need to show up at the SuiHub Lagos on the 27th. The judging will be happening virtually on the 28th.The ChallengePick a real problem — yours or someone else's. Build an agent or prompt that solves it using Walrus Memory. Write down the exact prompt you used. Show us it works with a short video. What makes a good submission? Anything that makes an agent more useful by giving it memory.That could be:- A personal assistant that remembers your preferences, routines, or goals- A coding agent that remembers architectural decisions and never re-asks for context- A research or study agent that recalls what it got wrong and improves next time- A meeting or relationship-tracking agent that remembers everything about the people you talk toSomething we haven't thought of — that's the one we're most excited to see While you can look at these and use as inspiration, we want to see something which is new and useful for you: https://www.deepsurge.xyz/projects?hackathon=77d00c13-63af-4bfa-b6a3-ce3fe5662e25The bar is simple: does it solve a real problem, does it work, and would you actually keep using it?Submission Requirements - The prompt: full copy-pasteable text of the exact prompt you used — ready for someone else to run and get the same result. This is the part we most want from you; it's what gets shared forward. - What you built: a short, plain-language explanation of what your agent or prompt does - What it remembers: what it stores, when, and why — what problem that memory solves - Proof it works: With links to stored memoryResources 📚 Walrus Memory resources: https://app.notion.com/p/mystenlabs/Walrus-Session-5-3756d9dcb4e9808ca16fc8c22562e3c6Judging Criteria Does it solve a real problem? A simple prompt solving something genuine beats something clever nobody would use Is it well-crafted? Does it give the agent clear rules for what to remember and when? Does it work? Can someone else pick it up? Is the shared prompt clear and complete enough that another person could run it themselves? Judged by the Walrus Foundation.Prizes