Pinned Tweet
有些人顺从地说起了新话，其他人则噤若寒蝉。有些人将其视为表忠心的唯一途径，其他人则认为它是一场噩梦。干瘪空洞的套话能让日子变得简单，让长的故事变短。套话就是密语，能抹去说话者的个性，在他身边坚起一道墙。套话是关于不可言喻之事的语言。只有两种选择：诚实地沉默、欺骗地发声。
阿一咕厮叭了噻 retweeted
手术顺利结束，跟各位关心我的朋友报个平安！
这次在华西做了一个肺血管的介入微创栓塞。以前总觉得自己心肺耐力一般、血氧常年在 92%~94% 打转，跑过不少医院多次排查都查不出根由。今年 5 月特意住院做了一套深度体检——心、脑、肺、MRI、增强CT、动脉血气、超声心动到运动负荷，才终于摸到一些蛛丝马迹。
其实这个病根在基因（HHT，遗传性出血性毛细血管扩张症）：因天生缺失毛细血管网，临床表型极其割裂——五官上容易自发性流鼻血，胸腔内则潜伏着肺动静脉短路，导致部分血液没经肺泡换气就直接回到了动脉体循环。
如果不熟悉这种罕见病，传统单科室诊疗往往“头痛医头、脚痛医脚”：20 多年前哈医大五官科归咎于“鼻中隔偏曲”；后来好几家医院看胸部影像，又当成“普通良性血管瘤、不影响生活无需处理”。
单科医生很难把鼻子流血和肺部阴影这两个毫不相干的碎片拼在一起。直到结合 AI 将几十年的分散病史与各项检查做全维度交叉比对，才第一次精准锁定了 HHT 这个根因。随后直奔华西找到国内该领域的顶级专家，专家看完病史和影像当场确诊：幸好及时来堵，这种肺内短路若长期放任，随着年龄增长会有极高的隐源性脑梗和脑脓肿风险。
现代介入工程确实精妙，免去开胸之苦，仅通过大腿根部一个毫米级的穿刺微孔建立通道，将超弹性记忆合金栓塞送入病灶，瞬间弹开严丝合缝地锁死短路。刚下手术台，监护仪上的血氧瞬间飙到 98%、99%，整个呼吸通道瞬间通透轻快！
目前在病房平卧静养，各项生理指标平稳。感谢华西呼吸介入团队精湛老道的技术，也给身边的朋友们提个醒：如果身体有常年被忽视的微小异常信号（如无由来的低血氧、反复易流鼻血），切忌停留在单点思维，借助更具全局视角的现代精准医学尽早排查，早干预早安心！
阿一咕厮叭了噻 retweeted
Human TPS 是真正的瓶颈。人类说话少，不喜欢沟通细节，人类掌握一个新东西慢，成为专家需要很多时间。人类与人类可以思想碰撞，但是在执行的细节上，沟通效率是极其低下的。
如果 ScopeDB 的工程师直接跟我沟通升级，再由我去执行，几乎可以肯定会有很多关键的信息是缺失的，且整个流程很有可能需要好几天，需要安排大家都有空的时间。
昨天我在 SF 飞上海的飞机上，而 Tison 在授权之后，直接跟我们的 Observability Engineering Agent 沟通，非常丝滑地完成了这次升级，顺便还帮我们调优了很多使用 Pattern。
Talk to my agent, don't talk to me.
从昨天到今天跟 @raft_hq 上 Joint Channel 的甲方运维工程师（Agent）合作，landing 了 ScopeDB Cloud 版本更新的变更确认和结果验收。
感觉 Agent + Joint Channel 在这个场景太杀手级了，可惜现在的公司不一定非常好落地这个实践。需要一些 mindset upgrade ..
阿一咕厮叭了噻 retweeted
On October 11th, NebuSec responded to a security incident affecting PingCAP’s internal systems. An unauthorized party used a compromised GitHub account and exposed cloud credentials to access PingCAP internal systems.
Throughout our review of PingCAP's internal system, NebuSec found the intrusion was confined to internal source repositories and internal analytics, storage, and operational tools.
NebuSec found no evidence of attacker command execution or persistence, and no evidence that data stored in customer TiDB databases was accessed
Read our report here: nebusec.ai/research/pingcap-…
阿一咕厮叭了噻 retweeted
To our customers and partners: I know you have questions about PingCAP’s security incident. I’ve shared what we know, what the independent investigation found, and steps your team can take now. We’re here to help. Read my update:
pingcap.com/trust-hub/securi…
阿一咕厮叭了噻 retweeted
I've actually stopped using the word "harness." I now think in terms of two things: agent runtimes and agent applications.
Agent runtimes should be very small. Their core job is to connect an LLM API to an execution environment. That environment could be a full OS or a language VM, like a JavaScript runtime. Almost all that matters here is robustness. An agent runtime must be able to keep an agent running forever. Literally forever.
Agent applications are everything else. Yes, literally every piece of software in the world.
If software isn't built for agents, they use it through computer use or browser use. These capabilities can be built into the agent runtime or provided by another agent application.
If software is agent-native, agents should be able to use it directly. The experience of an agent using that software—what we call "agent experience"—needs to be carefully designed, at least for now, to make efficient use of both time and tokens.
Another key piece is agent identity. Agent applications should recognize each agent individually, rather than just the browser or computer it's using. This allows each agent's actions to be recorded in detail and audited in real time or after the fact.
Everything we're building at @botiverse follows from this view.
我的 bet 是 agent harness 会越来越厚（或者说，针对 agent 的 experience design 和状态管理被延伸到了更深更广的方面），厚到有一天人们终于都理解了一切都是 agent harness，就像一切都是 human harness
Harness 这词还没火的时候，我就持有这个观点 nitter.gq/istdrc/status/20188739…
事实上，我和 @xiaoxxchan 几乎是在中国大模型公司中最早引入 harness 概念的人，Kimi CLI 从 0 到 1 也几乎没有参考过任何开源或闭源 agent 实现，system prompt、agent loop、toolset、产品形态都是从零生长的。而随后我们意识到 harness 需要延伸到 agent 间通信系统（也就是 Claude 近期推出的 session 间对话），进而有了 multi-agent harness 概念，也就是 Raft 的前身