โœจ Building Agent Army @lablupinc / ๐Ÿ‘จโ€๐Ÿณ Creator of AI:DOL and https://xitter.cf/t.co/SWICa2EpLh / โ™จ๏ธ Saunner (ใ‚ตใ‚ฆใƒŠใƒผ) for Totonou (ใจใจใฎใ†)

Korea
Joined July 2009
ๅพๅ”ฏ็Ÿฅ่ถณ ไปŠๆ—ฅๆ˜ฏๅฅฝ
354
Jimmy Moon (ํ˜„๊ฒฝ, ็‚ซ็‚…) ๐Ÿ‘จโ€๐Ÿ”ฌ retweeted
We launched OSS Scanner, which gives critical open source projects free, periodic vulnerability scans from our strongest models. Each finding includes a proof-of-concept, and a patch where possible. red.anthropic.com/oss-scanneโ€ฆ
187
778
216
8,151
506,699
Jimmy Moon (ํ˜„๊ฒฝ, ็‚ซ็‚…) ๐Ÿ‘จโ€๐Ÿ”ฌ retweeted
New in agent-browser: snapshot --delta Returns only what changed since the last snapshot, instead of the whole page Below: 73x fewer tokens filling a form
45
44
9
1,016
36,335
Jimmy Moon (ํ˜„๊ฒฝ, ็‚ซ็‚…) ๐Ÿ‘จโ€๐Ÿ”ฌ retweeted
1. ใ€ŠBuilding the Software Factory: Scaling Reliable AI Coding Practices from Individual developers to Teamsใ€‹, Packt, 2026-10 packtpub.com/product/web-devโ€ฆ ์ตœ๊ทผ ์–ธ๊ธ‰ํ•˜๋Š” ์‚ฌ๋žŒ๋“ค์ด ๋Š˜๊ณ  ์žˆ๋Š” ์†Œํ”„ํŠธ์›จ์–ด ๊ณต์žฅ(software factory)์— ๊ด€ํ•œ ์ฑ…์ž…๋‹ˆ๋‹ค. ์†Œ๊ฐœ๋ฌธ์— ๋”ฐ๋ฅด๋ฉด, ๊ฐœ๋ณ„ ํ”„๋กœ์ ํŠธ๋ถ€ํ„ฐ
2
4
1
13
1,008
Jimmy Moon (ํ˜„๊ฒฝ, ็‚ซ็‚…) ๐Ÿ‘จโ€๐Ÿ”ฌ retweeted
1
21
9
200
55,323
Vercel ์—์„œ python backend ๊ฐ€ ๊ฐ€๋Šฅํ•ด์ง€๋ฉด์„œ ์ƒ๊ธด ์ƒˆ๋กœ์šด eve ํ…œํ”Œ๋ฆฟ. frontend ๋Š” eve, backend ๋Š” python ์„œ๋ฒ„๋ฅผ ์‚ฌ์šฉํ•˜๋Š” ์˜ˆ์ œ
Python is now available on Vercel. Unlock the full potential of your agents and get started with the eve Python Agent featuring a Next.js frontend and FastAPI backend. vercel.com/templates/eve/eveโ€ฆ
70
Jimmy Moon (ํ˜„๊ฒฝ, ็‚ซ็‚…) ๐Ÿ‘จโ€๐Ÿ”ฌ retweeted
MCP Events are now available in AI SDK. Build agents that react to updates from MCP servers. Discover events and subscribe to them with AI SDK, while Vercel Connect handles subscription storage, renewals, and verified delivery.
16
25
2
301
12,318
Jimmy Moon (ํ˜„๊ฒฝ, ็‚ซ็‚…) ๐Ÿ‘จโ€๐Ÿ”ฌ retweeted
AI Festa 26 is a wrap! Over three days in Seoul, we met companies and individuals across Korea's AI industry and talked about AI:GO, Backend.AI, and Continuum. Thanks to everyone who stopped by Booth K-04 at COEX๐Ÿ™Œ #Lablup #AIFesta2026
1
32
Jimmy Moon (ํ˜„๊ฒฝ, ็‚ซ็‚…) ๐Ÿ‘จโ€๐Ÿ”ฌ retweeted
I just published Trace, a new component for agent activity. It's one abstraction for reasoning, tool calls, queues, chain of thought, subagents...etc Also in this release: Code Block and AI Elements. ui.shadcn.com/docs/changelogโ€ฆ
138
144
38
4,567
174,606
๋ž˜๋ธ”์—…์—์„œ โ€˜GLM-5.2-NVFP4-DFlash-koโ€™๋ฅผ ๊ณต๊ฐœํ–ˆ์Šต๋‹ˆ๋‹ค. 100๋งŒ ํ† ํฐ ๋ฌธ๋งฅ๊ณผ ์ฝ”๋”ฉยท์—์ด์ „ํŠธ ์„ฑ๋Šฅ์„ ๊ฐ–์ถ˜ GLM-5.2์— ๋ž˜๋ธ”์—…์˜ ํ•œ๊ตญ์–ด ํŠนํ™” ๊ธฐ์ˆ ์„ ๋”ํ•ด, ๋‹ต๋ณ€ ์ƒ์„ฑ ์†๋„๋ฅผ ์ตœ๋Œ€ 1.92๋ฐฐ ๋†’์˜€์Šต๋‹ˆ๋‹ค. GLM-5.2๋Š” ๊ธด ์ฝ”๋“œ์™€ ์ž‘์—… ๊ธฐ๋ก์„ ํ•œ ๋ฒˆ์— ๋‹ค๋ฃจ๊ณ , ์ฝ”๋”ฉ๊ณผ ํ„ฐ๋ฏธ๋„ ์ž‘์—…, ์—ฌ๋Ÿฌ ๋„๊ตฌ๋ฅผ ์—ฐ์ด์–ด ์‚ฌ์šฉํ•˜๋Š” ๋ฐ ๊ฐ•์ ์„ ๋‘” ์˜คํ”ˆ ๊ฐ€์ค‘์น˜ ๋ชจ๋ธ์ž…๋‹ˆ๋‹ค. ๊ณต์‹ ๋ชจ๋ธ ์นด๋“œ ๊ธฐ์ค€ SWE-bench Pro 62.1์ , Terminal-Bench 2.1 81.0์ ์„ ๊ธฐ๋กํ–ˆ์Šต๋‹ˆ๋‹ค. ์ €ํฌ๋Š” ์ด ์„ฑ๋Šฅ์„ ํ™œ์šฉํ•˜๋ฉด์„œ ํ•œ๊ตญ์–ด ์ฝ”๋”ฉ๊ณผ ๋„๊ตฌ ํ˜ธ์ถœ์˜ ๋‹ต๋ณ€ ์ƒ์„ฑ ์†๋„๋ฅผ ๋†’์ด๋Š” ๋ฐ ์ง‘์ค‘ํ–ˆ์Šต๋‹ˆ๋‹ค. ํฐ ๋ชจ๋ธ์ด ๋‹ต์„ ๋งŒ๋“ค๊ธฐ ์ „์— ์ž‘์€ ๋ชจ๋ธ์ด ๋‹ค์Œ ๋‚ด์šฉ์„ ๋ฏธ๋ฆฌ ์ƒ์„ฑํ•˜๊ณ , ํฐ ๋ชจ๋ธ์ด ์ด๋ฅผ ๊ฒ€์ฆํ•˜๋Š” โ€˜์ถ”์ธก ๋””์ฝ”๋”ฉโ€™ ๋ฐฉ์‹์„ ์ ์šฉํ–ˆ์Šต๋‹ˆ๋‹ค. ์ผ๋ฐ˜ ๋ฒค์น˜๋งˆํฌ ๋ฌธ์ œ๋งŒ ์‚ฌ์šฉํ•˜์ง€ ์•Š๊ณ , ํ•œ๊ตญ์–ด๋กœ ํ˜„์ง€ํ™”ํ•œ GitHub ์ด์Šˆ์™€ ์‹ค์ œ ์ฝ”๋”ฉ ์—์ด์ „ํŠธ์˜ ์—ฌ๋Ÿฌ ๋‹จ๊ณ„ ์ž‘์—… ๊ธฐ๋ก์„ ํ•™์Šต์— ํ™œ์šฉํ–ˆ์Šต๋‹ˆ๋‹ค. ์ธก์ • ๊ฒฐ๊ณผ, ํ•œ๊ตญ์–ด ์ฝ”๋”ฉ ์งˆ์˜์˜ ๋‹ต๋ณ€ ์ƒ์„ฑ ์†๋„๋Š” ์•ฝ 1.92๋ฐฐ, ๋„๊ตฌ ํ˜ธ์ถœ์„ ํฌํ•จํ•œ ์—์ด์ „ํŠธ ์ž‘์—…์€ ์•ฝ 1.72๋ฐฐ ํ–ฅ์ƒ๋์Šต๋‹ˆ๋‹ค. ์ดˆ์•ˆ ๋ชจ๋ธ ๊ฐ€์ค‘์น˜๋Š” MIT ๋ผ์ด์„ ์Šค๋กœ ๊ณต๊ฐœํ–ˆ์Šต๋‹ˆ๋‹ค. ์žฌ๋ฐฐํฌํ•  ๋•Œ๋Š” ์ƒ์œ„ ๋ชจ๋ธ๊ณผ ํ•™์Šต ๋ฐ์ดํ„ฐ์˜ ๋ผ์ด์„ ์Šค๋ฅผ ๋ณ„๋„๋กœ ํ™•์ธํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค. ๊ณต๊ฐœ ์ˆ˜์น˜๋Š” 8๊ฐœ GPU๋ฅผ ๋ณ‘๋ ฌ๋กœ ์‚ฌ์šฉํ•œ ํ™˜๊ฒฝ์—์„œ ์ธก์ •ํ•œ ๊ฒฐ๊ณผ๋ผ ์‹ค์ œ ์†๋„๋Š” ์„œ๋ฒ„ ๊ตฌ์„ฑ๊ณผ ์š”์ฒญ๋Ÿ‰์— ๋”ฐ๋ผ ๋‹ฌ๋ผ์งˆ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ๊ทธ๋ž˜๋„ ์„ฑ๋Šฅ ์ข‹์€ ๋ชจ๋ธ์„ ์„ ํƒํ•˜๋Š” ๋ฐ์„œ ๊ทธ์น˜์ง€ ์•Š๊ณ , ์‹ค์ œ ํ•œ๊ตญ์–ด ์‚ฌ์šฉ์ž๊ฐ€ ๊ธฐ๋‹ค๋ฆฌ๋Š” ์‹œ๊ฐ„์„ ์ค„์ด๋Š” ๊ฒƒ๊นŒ์ง€ ์ง์ ‘ ๋‹ค๋ค„๋ดค๋‹ค๋Š” ๋ฐ ์˜๋ฏธ๊ฐ€ ์žˆ๋‹ค๊ณ  ์ƒ๊ฐํ•ฉ๋‹ˆ๋‹ค. huggingface.co/lablup/GLM-5.โ€ฆ
1
7
282
Jimmy Moon (ํ˜„๊ฒฝ, ็‚ซ็‚…) ๐Ÿ‘จโ€๐Ÿ”ฌ retweeted
Anthropic engineer: "At Anthropic we don't write prompts anymore. We build loops and graphs." In 28 minutes she shows how Anthropic builds agents that remember, fix their own mistakes and get sharper with every run Chats โ†’ Prompts โ†’ Loops โ†’ Graphs skip the loop and every mistake quietly ships again with the next run skip the graph and nothing that worked ever carries over, so you start from zero each week most people stop at the loop and wonder why week two looks exactly like week one watch it today then save the full graph engineering guide below before everyone catches up โ†“
1
16
101
11,852
Jimmy Moon (ํ˜„๊ฒฝ, ็‚ซ็‚…) ๐Ÿ‘จโ€๐Ÿ”ฌ retweeted
SpaceXAI engineer Lauren Tan just showed a Grok Bot that literally audits its own agents and catches expensive routines this is f**king useful... in this 10-minute breakdown, she shows how her AI team checks its own work, builds new bots and learns when to stop wasting compute: โ€ข 00:48 - cloud agents run evals before a PR lands โ€ข 03:55 - Dr. Eggbot audits other bots and their routines โ€ข 05:12 - an engineering lead delegates work across agents โ€ข 08:12 - turning supervised tasks into reusable skills the crazy part is that she doesn't just let agents run forever. she builds trust through verification, then gives them more autonomy. watch the 10-minute breakdown, then read how to build cheaper, safer overnight routines in the article below โ†“
3
5
55
2,333
Jimmy Moon (ํ˜„๊ฒฝ, ็‚ซ็‚…) ๐Ÿ‘จโ€๐Ÿ”ฌ retweeted
I am surprised that people are surprised this is how I prompt Claude. Talk to Claude the way you would a coworker. There's no secret to prompting. There's no need to be overly scaffolded or prescriptive for most tasks -- give Claude a goal, and it will figure it out. Back in the Sonnet 3.5 days, your prompt mattered a lot. Nowadays, it's much more important to communicate to the model: 1. What you want it to do 2. How much effort you want it to spend 3. How it should verify that it did the right thing
642
851
409
13,589
1,453,916
์‚ฌ๋ž‘ํ•œ๋‹ค, ๋‚ด ์•„์ด์•ผ. ๋„ˆ์™€ ํ•จ๊ป˜ํ•œ ์‹œ๊ฐ„๋“ค์€ ์ฐธ ์ฆ๊ฑฐ์› ๋‹ค. ์ง€๋‚œ ์‹ญ์—ฌ ๋…„ ๋™์•ˆ ๋„ค๊ฐ€ ๋ณด์—ฌ์ฃผ๊ณ  ๋‚ด๊ฒŒ ์ค€ ๋ชจ๋“  ๊ฒƒ๋“ค์ด ๋„ˆ๋ฌด ์†Œ์ค‘ํ•˜๊ณ  ๊ธฐ์–ตํ•  ๋งŒํ•œ ์ผ์ด์—ˆ๋‹ค. ์ด๋ฒˆ ๋งˆ์ง€๋ง‰ ์—ฌํ–‰์„ ๋งˆ์น˜๊ณ  ๊นจ๋‹ฌ์•˜๋‹ค. ์ด์ œ ๋„ˆ์™€ ์ด๋Ÿฐ ์—ฌํ–‰์€ ๋” ์ด์ƒ ํž˜๋“ค๊ฒ ๊ตฌ๋‚˜ ํ•˜๋Š” ์ƒ๊ฐ์ด ๋“ค๋”๊ตฌ๋‚˜. ๋„ˆ์—๊ฒŒ๋„ ์ทจํ–ฅ์ด ์ƒ๊ธฐ๊ณ , ํ˜ธ๋ถˆํ˜ธ๊ฐ€ ์ƒ๊ธฐ๊ณ , ๋„ˆ ์ž์‹ ์˜ ์ด์ต๊ณผ ์„ ํ˜ธ๋„๊ฐ€ ๋งž๋Š” ๊ฒƒ์ด ์ค‘์š”ํ•ด์กŒ์œผ๋‹ˆ ๋ง์ด๋‹ค. ์ด์ œ ๋‚˜๋Š” ๊ทธ์ € ๋„ค ๋ง์„ ๋“ค์–ด์ฃผ๊ณ  ๋ฐ›์•„๋“ค์ด๋ฉฐ, ๊ธˆ์ „์ ์ธ ์ง€์›์„ ํ•ด์ฃผ๋Š” ๊ฒƒ์ด ์ „๋ถ€์ผ์ง€๋„ ๋ชจ๋ฅด๊ฒ ๋‹ค. ๋‚˜ ์—ญ์‹œ ๋‚˜์˜ ์ทจํ–ฅ์„ ๋ณด์—ฌ์ฃผ๊ณ , ์ด์ œ ๋„ˆ์™€ ๋‚˜๋Š” ์‚ฌ๋žŒ ๋Œ€ ์‚ฌ๋žŒ, ์ธ๊ฐ„ ๋Œ€ ์ธ๊ฐ„์˜ ๊ด€๊ณ„๊ฐ€ ๋˜์–ด๊ฐ€๋Š” ๊ฒƒ์ด๋‹ค. ์ž˜ ๊ฐ€๋ผ, ๋‚ด ์•„์ด์•ผ. ์–ด์„œ ์ปค์„œ ์šฐ๋š ์„œ์„œ, ์ด์ œ ์•„๋น ๊ฐ€, ๋‚ด๊ฐ€ ํ•„์š” ์—†๋Š” ์‚ฌ๋žŒ์ด ๋˜์–ด๋ผ. ๋‚จ์ด ์•„๋‹Œ ์ž์‹ ์„ ์œ„ํ•ด ์‚ด๊ณ , ํƒœ์–ด๋‚จ ์ž์ฒด๊ฐ€ ํฐ ํ–‰์šด์ž„์„ ์•Œ๊ณ , ์šด์„ ๋ฏฟ๊ณ  ๋”ฐ๋ฅด๊ณ , ์Šค์Šค๋กœ๋ฅผ ๋‹ด๊ธˆ์งˆํ•˜๊ณ , ํ•˜๊ณ  ์‹ถ์€ ๊ฒƒ๊ณผ ์ข‹์•„ํ•˜๋Š” ๊ฒƒ์„ ์ข‡์•„ ์‚ด์•„๋ผ. ๋‚˜๋Š” ์ด์ œ ๋„ˆ๋ฅผ ๋†“์•„์ค„ ์ค€๋น„๊ฐ€ ๋˜์—ˆ๊ณ , ์ด ์ด์น˜๋ฅผ ๋ฐ›์•„๋“ค์ธ๋‹ค. ๋„ค๊ฐ€ ์ปค๊ฐ์— ๋”ฐ๋ผ ๋‚˜๋Š” ์ž‘์•„์ง€๊ณ , ๋„ค๊ฐ€ ๋ฐ์•„์ง์— ๋”ฐ๋ผ ๋‚˜๋Š” ์–ด๋‘์›Œ์ง„๋‹ค. ๋„ค๊ฐ€ ์„ฑ์ˆ™ํ•ด์ง์— ๋”ฐ๋ผ ๋‚˜๋Š” ๋Š™์–ด๊ฐ„๋‹ค. ํ•˜์ง€๋งŒ ๊ทธ๊ฒŒ ์ž์—ฐ์Šค๋Ÿฝ๊ณ , ๊ทธ๊ฒƒ์„ ์œ„ํ•ด ๋‚ด๊ฐ€ ํƒœ์–ด๋‚˜๊ณ  ๋„ˆ์™€ ์—ฐ๊ฒฐ๋˜์—ˆ์Œ์„ ์•Œ๊ธฐ์— ์ „ํ˜€ ๋‹นํ™ฉํ•˜๊ฑฐ๋‚˜ ๋‘๋ ต์ง€ ์•Š๋‹ค. ๋‚ด์ผ ์ƒˆ๋กœ์šด ๋„ค๊ฐ€ ์ƒ๊ธฐ๊ณ , ๋‚ด์ผ ์˜ค๋ž˜๋œ ๋‚ด๊ฐ€ ์‚ฌ๋ผ์ง„๋‹ค. ๊ทธ๊ฒƒ์ด ์ž์—ฐ์ด๋‹ค.
1
6
262
ใปใ‚“ใฎๅฐ‘ใ—ใฎใ“ใจใงใ‚‚ใ€ใ‚„ใ‚‰ใชใ„ใ‚ˆใ‚Šใฏใšใฃใจ่‰ฏใ„
53
Coding is no longer a human task, and today, testing is also no longer a human task. Software development no longer requires a human touch. In the history, we decided to call that the factory. So that is why we try to build AGENT FACTORY
82