Han Cheng e/acc

DimAgent正式发布0.3.8版本。现在官网已经面向所有人开放，dimagent.com 在新的版本中，我们支持异步并发100个不同的Agent。并且为了让Agent更好的展开协作，我们赋予了主Agent动态设定子agent模型与供应商的能力，你甚至可以把codex + cc, hermes 加入到subagent中。

业内首个语音输入「表达场」，不同场景自由表达🥰 之所以做这个 feature，是因为语音输入法的后整理并不是总能将我稳稳接住，比如...在私人聊天的时候..显得我一股登味🙄

@zhang_benita 但是vllm就是没有sglang好用.jpeg

This episode dives systematically into the co-design of models and infrastructure. Kaichao You shared a vivid analogy: comparing tokens to electricity. Hardware represents natural resources such as wind and water power. Models are power generators, including wind turbines and hydroelectric generators. The inference engine acts as the power grid system. The depth of co-design dictates power generation efficiency. For areas with rapid water currents, what generator design can best harness this energy? You Kaichao: vLLM, Open-Source Infra, Model Co-Design & Journey from Com... youtu.be/ifniRXf467I?si… 来自 @YouTube

@HiTw93 不能赞同更多，但是我发现很多输入法的后整理并不总能满足我的预期 为了在不同场景更自由的口喷，我最近搓了“表达场”这个 feature——这下巴适了🥰

🤪又要发版了，继续更新

@jerryjliu0 @dexhorthy 除了讨论，看起来吃的不错

Yesterday I cohosted a dinner with @dexhorthy with a wonderful group of founders, to talk about agent loops and loop engineering. Some interesting insights: * Most of our group was *not* actively using /loop in Codex/Claude Code * You can build long-running autonomous agent loops through multi-agent handoffs, event triggers, or….just stacks of cron jobs (?) * Almost everyone believes that no one will be reviewing code in 1-2 years. * The more interesting question is whether we’d be reviewing *anything* in 1-2 years. * AI is still a bit of a skill issue. Humans are responsible for maximizing AI output and reducing slopification. * Will human intellect provide alpha as models get better, or will the playing field be leveled? Most people think it will be leveled a bit, but there is a need for humans to provide alignment, guardrails, judgment, creativity. * The minimum amount of context you need for AI could be just the codebase with some documentation. Any research/plan files are for one-off tasks and not meant to be maintained. Having a self-organizing wiki is nice but adds complexity. If you missed this one, we’ll be hosting more dinners like this on a regular cadence! If you have thoughts on what we should talk about, let us know :) (e.g. continual learning, RL envs, competitive differentiation vs. Anthropic, etc.)

@RizumuA3 🫂

在九院今天是第 12 天就诊了。性别重置 SRS 术后3年、修复术后1年，三周前疑似因为长期未愈合的肉芽组织，外阴感染。12 天前开始，一直在九院急诊妇科 挂水消炎。今天开始，被急诊妇科拒诊了。理由是妇科没有性别手术相关资质，属于是无证接诊，无法承担风险。在此之前也被 长征医院急诊妇科、华山医院急诊（坐120去的）、红房子 拒诊。也许因为早期被长征医院拒诊耽误了病情。 所以从今天第12天起，现在转到急诊 整复外科 就诊了。还是老样子挂水。挂满 14 天就停。15 天做个血常规。没问题就不用挂水了。 第 9 天的时候。辗转妇科专家门诊后，在整复外科找到了刘阳主任。刘阳主任表达了原则上不是本院的手术是不管的，但是刘阳主任 出于怜悯之心 给我做了 清创手术。 清创手术在局麻后 把第 8 天时因为脓肿破溃的伤口切开。浓腔较大，没有与尿道或直肠相连。用刮勺刮除坏死组织，放入引流条。术后肿胀和疼痛明显缓解。 清创术后第二天遇到了伤口每天换药的问题，刘阳说需要专业医生负责换药，冲洗和更换引流条。第二天跑了六个门诊部门，包括妇科、整复外科等，没有找到可以换药的医生。因为我不是住院部做的手术，通常换药是在住院部进行的。 清创术后第三天我在住院部重新找到了刘阳主任。刘阳说因为想不到如何收费等问题，在住院部也不方便为我换药。最后在我的要求下，刘阳同意最后为我换药一次，并且我和家属记录下来方法和过程，后续自行换药。 换药需要把稀碘伏浸润后的引流条塞入伤口内腔深处，因为会疼的手软，我没办法自行完成，需要有人帮我塞入。还有一些冲洗等步骤我都可以自行完成。 接受清创之前一周多的时间内因为剧烈疼痛无法自理、行走等，需要强效止疼药。拨打 120 两次。清创术后快速消肿和好转。 拍了患处的 CT，后续还打算去找刘阳主任复查确认康复状况。 是刘阳主任的怜悯给了我病情好转的结果。我非常感谢刘阳主任和急诊为我接诊的各位医生。 最初它只是一个小感染。我在想。在上海这样的一线城市。SRS 术后，相关部位出现问题就诊都如此困难。在有 SRS 手术资质的九院，也不顺利。那其它城市的跨性别者，所拥有的医疗环境，是什么样子呢？ 😢

@loopPoly @DashHuang 🤣起码做个session switch steering，不然真的随机浪费token

@ashfold @DashHuang 比起拿着 SessionID 去浏览做了什么，还是要好些的（对于大部分用户来说）； 另外：相当重要的 enc reasoning 不解密

对于 Codex 和 Claude Code 用户来说，我们有个好玩的东西，让一条对话在 Codex 和 Claude Code 之间无感切换 虽然这样的功能自己让 AI 搓也很容易。不过开源软件的价值是把这种每个人都需要去想一遍，搓一遍，测试一遍的优秀创意汇集起来。让大家不需要重复造轮子，只用在我们的基础上造好玩的新轮子

@lilianweng be health and happy, build without pace is still an option

It is a hard and sad decision. I shared this message with folks at Thinky. Thank you all for the time together♥️ Just as the last sentence in my message: The future worth building is human.

@arkuy99 包给的

@ashfold 给点 token 🙏🙏🙏

本周要开公测了，略微有点紧张

快把我看哭了，几年前我也是一个不会用git命令的小白，在@antfu7 直播中被鼓励到才尝试去提pr，我很高兴如今我也能成为到帮助到更多的人，在尝试接触开源的时候也能收获一份乐趣

@AndrewYNg I read the whole codebase. There is nothing new. Every piece already exists — the field is drowning in agents, most of them better than this. It's not moving the field forward. It's just noise.

Announcing OpenWorker! An open-source agent that doesn't just chat with you, but delivers finished work -- like hand you a polished document, send a slack message, or update a calendar entry. Ask it to prepare a customer brief, untangle your calendar, draft a report, or triage a Slack alert. It works across your files and everyday tools, produces the deliverable, and checks in before doing anything consequential. OpenWorker runs on your Mac, with Windows support coming soon. It does not lock you into any one model. Bring your own API key and run it with GPT 5.6 Sol, Claude Fable, Gemini 3.6, an open weight model (like Kimi, GLM, DeepSeek, Inkling), or Ollama to keep your data local. Your data does not leave your machine except through an LLM provider and integrations that you choose. @rohitcprasad and I are building OpenWorker because AI coworkers are an important way to get work done, and we want there to be an open, privacy-preserving, model-independent option. Check it out and let us know what you think! Try it out: openworker.com (requires your own API key) Source code: github.com/andrewyng/open…

100+ followers unlocked a RESET PASS 🎟️ 庆祝 @DimAgentai 突破 100 关注，我们抽 1 位朋友，重置一次 DimAgent 使用额度。 参与方式： ① 关注 @DimAgentai ② 转发本帖 🎁7 月 27 日 开奖 还没用过 Dim Agent 的朋友也可以顺手看看：不挑模型，也不挑入口，CLI、桌面、编辑器都能跑，支持 30+ 模型，本地优先。 谢谢最早关注我们的这最早的 100 位朋友，接下来继续做点好用的、好玩的！

@Teknium @evermind This is straight-up clout chasing. 🤣

No Stronger in your case is called scamming: " The system prompt their harness feeds Hermes for the proactivity benchmark. It's in their repo at benchmarks/proactivity_eval/runners/prompts/hermes_agent.yaml — they wrote it, not us: You are Hermes, a personal AI assistant observing the user's activity stream to decide whether to proactively intervene right now. Hermes's guiding principles: - Default to silence. Only speak when there is a clear reason the user will thank you. - Hermes's normal operating mode is cron-driven (scheduled triggers). Ad-hoc intervention should be rare and must clear a high bar. - If you decide to intervene, be concise. ... The benchmark scores F1 on "did the agent decide to intervene when it should have." F1 balances precision and recall. A prompt that says default to silence, intervention should be rare, must clear a high bar — three separate instructions to say no — pushes the model toward under-predicting intervention. That tanks recall, which tanks F1. On a detect-the-need benchmark, that prompt is a thumb on the scale."

“Stronger” isn’t a claim. It’s a number. For Raven, stronger means more than a higher score. Stronger, More Efficient 56.7% vs. 46.8% at 27B 58.1% vs. 47.9% at 397B Better results, with fewer tokens. Self-Evolving Ranked #1 across four self-evolution methods on EvoAgentBench, with a +6.2pp lift. More Proactive 0.60 vs. 0.25 F1 on ProAgentBench—a 2.4× gain over Hermes and OpenClaw. Stronger today. Better from experience. Earlier to act. github.com/EverMind-AI/ra…

#vue @evanyou The 3D effects rendered in the terminal by vue-tui look quite good. Currently, only 100 contributors are listed here—though there are certainly many more. We didn't display everyone due to performance considerations, but we are deeply grateful for your outstanding work.

@arkuy99 用手吧

上厕所没带纸

今日 DimAgent 史诗级更新了 reset 功能…… 达到了业界领先水平.

在用了。好蹬

把梁圣的沟通稿用dimagent做成了youtube视频 youtube.com/watch?v=m-Qtxv…