This is the list of winners from @evermind
's hackathon competition. Every submission receives a participation award worth $100, plus a small souvenir. In the end, over 400 people signed up, and we received 73 submissions. Many of the works were bursting with imagination,
@elliotchen100
-
Evermind Hackathon Winners: 400+ Participants, 73 Creative AI Submissions
By
–
-
Claude Design Launch Sparks Design Philosophy Debate
By
–
Claude Design 说: output 现在免费。 Karri 说: output 从来不是 design。 这个有点意思，这两天 Anthropic 发了 Claude Design。 同一时间，Linear 的 Karri Saarinen 发了一篇文章，标题叫「Output isn't design」。 哥们之前在，Airbnb， 然后那套设计语言(包括 Cereal 字体,Lottie
-
Claude Design: AI-Powered Visual Understanding for UI Creation
By
–
Claude Design 出了，读了一遍 blog，有个细节挺值得聊的。
— 艾略特 (@elliotchen100) 18 avril 2026
这东西底层跑的是 Opus 4.7，是 vision model。
跟 v0 或者 lovable 那些工具最大的区别不是功能多少，是出发点不同。大部分 AI 设计工具的思路是生成代码，代码渲染出 UI。Claude Design 的思路是先「看懂」你要什么，再去做。… https://t.co/RQhpFtNjOhClaude Design 出了，读了一遍 blog，有个细节挺值得聊的。 这东西底层跑的是 Opus 4.7，是 vision model。 跟 v0 或者 lovable 那些工具最大的区别不是功能多少，是出发点不同。大部分 AI 设计工具的思路是生成代码，代码渲染出 UI。Claude Design 的思路是先「看懂」你要什么，再去做。
-
Company Accepting Resumes for AI Positions
By
–
Thanks for testing, we'll fix it ASAP. Also, you can submit your resume
-
EverOS Open-Source Library Launch with Methods and Benchmarks
By
–
I'm thrilled to announce that the update we've been polishing for ages and ages has finally arrived—not just the product, but also Methods, Benchmarks, Use Cases, and the official website, and more. Specific updates: 1. EverOS – A fresh approach, an open-source library for
-
Memory Sparse Attention GitHub Repository Reaches 3000 Stars
By
–
讲真，我们没有想到 Memory Sparse Attention 的 GitHub Repo 会这么快到 3000 星。更没有想到大家的关注度这么高。也许真的印证了那句话吧，「能带走的才是记忆，带不走的则是过去」。
— 艾略特 (@elliotchen100) 12 avril 2026
起初，MSA 只是一篇论文。一个支持 1 亿 Token 上下文的端到端方案，大胆，也优雅。
核心就两个点：一个是 Dual… pic.twitter.com/DUBidTcl8rTo be honest, we didn't expect the GitHub Repo for Memory Sparse Attention to hit 3000 stars so quickly. Even less did we expect such high attention from everyone. Maybe it really proves that saying: "What you can take with you is memory; what you can't take is the past." At the
-
EverMind Launches Agent Capability Benchmark Test
By
–
稍微剧透一下，
@EverMind 马上会推出一个 Benchmark，可测试 Agent 的能力。 诸位的 OpenClaw 和 Hermes Agent 都可以测测看。 -
Vertical-Domain Agents Development Progresses Rapidly
By
–
Progress is very fast; from my experience, the development of vertical-domain Agents feels several levels higher than last year.
-
Hyperedges Package Multiple Elements Beyond Pairwise Graph Connections
By
–
That's a very good question. Actually, it's not entirely about relying on relevance to connect things. The hyperedge itself packages multiple related elements into a cohesive whole, which is different from ordinary graph memory that only makes pairwise connections—the packaging
-
GBrain: Open-Source Agent Using Markdown Knowledge Base
By
–
Garry Tan turned his 10,000+ markdown files into the brain of an Agent and open-sourced it, calling it GBrain—pretty interesting stuff. I did a quick AI-assisted dive into the code. The core idea: One markdown page per person, one per company. The top half lays out your most