avatar
0xFunky
20,099 Twitter followers
Follow
BioMedical AI Start up - CTO / Kaggle Competition - Master / @GooseCityDAO / Founder @codechainAI
Posts
avatar
0xFunky
04-01
Thread
Claude Code Buddy is now online, and you can raise pets! It's pretty funny. I was just thinking a while ago that constantly talking to Coding Agents was getting boring, so I made my own Coding Tamagotchi game. I didn't expect Claude Code to create one natively! 😂 The pet element images are all automatically generated by AI, including background removal and action sheets (I wrote a skill sheet for the agent to handle automatically). It assigns different values and generates different pet styles and evolution paths based on the session state. Encountering errors in the middle of coding will trigger a short battle. The more sessions you have, the more XP you accumulate for evolution. The pet has a memory function; it remembers what you did in the session and will remind you or say some silly things. However, this doesn't really help with the coding session. It's basically a pet system that automatically talks to you and raises you automatically. Actually, I had a lot of fun playing it, especially since I wrote a skill that can automatically generate these 2D pet element images. Every time I open a session, I'm really looking forward to seeing what kind of monsters will spawn. Maybe it's because I was a Pokémon and Digimon fan when I was a kid, so I've always dreamed of making my own 2D pixel-style pet companion game.
Oikon
@oikon48
04-01
Claude Code 公式エイプリルフール `/buddy` コマンドが入っています。 /buddy でコンパニオンが産まれて、チャット欄の横にいてくれます(Kawaii!)。ユーザーによって産まれるコンパニオンが違うので、良かったら何が出たか教えてください。
avatar
0xFunky
03-12
Thread
Inspired by Karpathy's autoresearch, I taught VibeHQ to evolve itself—not just a single agent, but the entire multi-agent collaboration mechanism. Seven fully automated runs with zero human intervention: • Token usage: 7.2M → 5.7M (peak value reduced by 62%) • Reduction in coordination-related issues (duplication of work, etc.): 4 → 0 • PM token waste: -91% Loop: benchmark → collaboration quantification and LLM analysis failure modes → /optimize-protocol rewrite coordination code → rebuild → repeat. The AI observes agent team collaboration failures, analyzes why they failed, and then modifies its own source code to coordinate the collaboration logic—all without human intervention. The AI completely organizes its own team's synergy. I looked at some related materials. Autoresearch is automatically optimizing model training. Ralph was a single-agent autonomous loop, and Gastown ran 20-30 Claude Codes simultaneously for orchestration, but it didn't have the ability to evolve. These were all very powerful, but later they all focused on evolving the capabilities of single agents. No one is evolving teamwork itself—how to divide tasks, how to avoid conflict, how to share context, how to unblock each other. Just like in the real world, AI teams also need time to gel. Imagine what this will become as it runs: • Agents develop their own team culture and working synergy. • It adapts to projects, allocating 3-person or 7-person teams based on the project's development progress. • The more projects done together, the stronger the team. • Agents can onboard new teammates during a project, automatically reassigning tasks. Honestly, what will it ultimately evolve into? I don't know, but that's actually the most exciting part. twitter.com/0x0funky/status/20...
avatar
0xFunky
03-10
I've recently been researching the Everything project, which integrates perpetual contracts, spot trading, prediction markets, and on-chain payments into a single account (TG). Its goal is quite straightforward: to make the trading experience, which was previously geared towards crypto natives, more user-friendly and accessible to the general public. It avoids complex operational processes and aims to reach a wider user base. Furthermore, thanks to OpenClaw, more people are using TG. I think the key to this Everything Public Sale isn't just about $E, but also about maximizing your Everything Points. If you're aiming for later airdrops or early-stage benefits, Phase 1 is essentially the most efficient period. The key points are simple: • Phase 1 offers a 3x points bonus. • Using $H increases points efficiency compared to USDT. • You can further stack points based on holdings/asset snapshots/trading volume/referrals. So this isn't simply about "buying early is cheaper," but rather, the earlier you enter, the lower the price, the higher the multiplier, and the longer you can accumulate points. My personal suggestion is this: If you already intend to participate, the most important thing isn't waiting, but maximizing points efficiency first. The most straightforward approach right now is: • Try to enter during Phase 1. • If you have $H, prioritize using $H. • Don't just hold after buying; make sure you get the holding bonus. • If you already trade, trading volume can help you stack up your points. • Referral programs can be an extra source of points. Later on, it's not just prices that will increase; even the point multiplier will decrease. With the same amount of capital, those who enter later will get fewer points. Therefore, if you're looking at points/airdrops/early entry points, I think Phase 1 is the most important area to focus on. In short: Entering earlier not only means buying cheaper, but also maximizing the efficiency of your Everything Points with the same cost. Use this link to log in directly and participate in the public sale: t.me/EverythingTradeOfficial_b...… Always DYOR.
Everything
@trdEverything
03-10
⚡️ Everything Phase 1 is officially LIVE. Welcome to your new home for trading. → Trade Perpetual futures with up to 1000x leverage → Earn 3× Everything Points for your airdrop allocation → $E Public Sale starts at $0.01 per $E ($50M FDV) → No KYC Don’t miss the start ↓
ETH
0.47%
loading indicator
Loading..