Chaos

🚨 Warning to Web3 developers, builders, artists, and contributors. My X account, GitHub, and several social accounts were compromised after a fake Web3 recruitment process. The project involved in this incident is LumaPad — @LumaPad_pro. I am writing this thread to document the full timeline: how trust was built, how I was pushed into a fake StreamYard meeting, how my machine was infected with Trojan:Win32/NetSupportRat!AMTB, and how that led to my X, Instagram, GitHub, and important development data being compromised. This is not drama. This is an incident report and a warning to the community.

I spent an entire day coding this game, so it’s disappointing to see it go unnoticed, especially when many AI-generated works were selected. I’d really appreciate it if you could take a closer look at my submission and give it another consideration. @Elyziumaa

Recently, a few security-related situations have made me more interested in how infrastructure monitoring could be improved. One thing I've been thinking about is the gap between detecting suspicious activity and actually responding to it. We already have plenty of tools for collecting logs, monitoring systems, and generating alerts. But investigating incidents, connecting events across different services, and deciding how to respond still require significant manual effort. I'm starting to explore whether a locally hosted AI agent could help bridge that gap. The idea is to combine traditional SIEM capabilities with an agent that can investigate events and interact with infrastructure through controlled interfaces. Not just reading logs or sending alerts, but potentially taking defensive actions such as temporarily blocking suspicious connections, revoking sessions, or isolating endpoints. There are several questions I want to investigate: How reliable can event correlation be with a small local model? How should the system distinguish suspicious behavior from legitimate administrative activity? And how much autonomy can an agent safely have without creating additional risks? Data privacy is another important constraint. Ideally, the entire system would operate locally, without sending logs, credentials, or infrastructure details to external AI services. For now, this is an early research direction rather than a finished project. I plan to experiment with detection rules, local LLMs, permission boundaries, and controlled response mechanisms in a lab environment. I'm particularly interested in whether a small, self-hosted system could provide meaningful security monitoring and response capabilities without the complexity of a full SOC deployment.

Spent all day bringing this game to life! 🎮 @Elyziumaa fam, come join the fun! LINK GAME : elyziuma-game.vercel.app/

Haha I want 1 duck

3/3 The five funds start with the same rules, but their holders can take them in different directions through weekly votes. That’s the part I’m most curious to watch: how differently will people manage the same starting ingredients? For this piece, I wanted the two characters busy preparing something together. One working the stove, the other finishing the cake. A fitting scene for a project built around shared decisions. The kitchen is open. Let’s see what we cook up. 🪑🍰 $TMF themutual.fun/prospectus