Learn AI (@LearnAI_MJ) header
Learn AI (@LearnAI_MJ) profile picture

AI enthusiast and AI artist. Eager to contribute to and grow the vibrant AI art community ๐Ÿ’™โค๏ธ๐Ÿค–๐Ÿฆพ

Joined August 2021
Photo by Learn AI (@LearnAI_MJ)Photo by Learn AI (@LearnAI_MJ)Photo by Learn AI (@LearnAI_MJ)Photo by Learn AI (@LearnAI_MJ)Photo by Learn AI (@LearnAI_MJ)Photo by Learn AI (@LearnAI_MJ)Photo by Learn AI (@LearnAI_MJ)Photo by Learn AI (@LearnAI_MJ)Photo by Learn AI (@LearnAI_MJ)Photo by Learn AI (@LearnAI_MJ)
Learn AI retweeted
ๅค–็ฝ‘ๆœ€่ฟ‘ๅพˆ็ซ็š„ไธ€ไธชๅญฆไน ๆ–นๆณ•๏ผšMIT ๅญฆ้œธ็”จ AI๏ผŒๅœจ48ๅฐๆ—ถๅ†…ๅฟซ้€Ÿๅƒ้€ไธ€้—จ้™Œ็”Ÿๅญฆ็ง‘ใ€‚ ็ฌฌไธ€ๆญฅไธๆ˜ฏ็›ดๆŽฅ้—ฎ ChatGPTโ€œ็ป™ๆˆ‘ๆ€ป็ป“ไธ€ไธ‹โ€๏ผŒ่€Œๆ˜ฏๅ…ˆๅปบ็ซ‹ไธ€ไธช่ถณๅคŸๅฏ้ ็š„ AI ็Ÿฅ่ฏ†ๅบ“ใ€‚ ไป–ไธŠไผ ไบ†6ๆœฌ่ฏฅ้ข†ๅŸŸ็š„็ปๅ…ธๆ•™็ง‘ไนฆใ€15็ฏ‡็ ”็ฉถ่ฎบๆ–‡๏ผŒไปฅๅŠ่ฟ™้—จๅญฆ็ง‘็š„ๅ…จ้ƒจ่ฏพๅ ‚่ฎฒไน‰ใ€‚ ่ฟ™ๆ ท AI ็š„ๅ›ž็ญ”ๅฐฑไธๅ†ๅชๆ˜ฏ็ฝ‘ไธŠไธœๆ‹ผ่ฅฟๅ‡‘๏ผŒ่€Œๆ˜ฏๅŸบไบŽๆŒ‡ๅฎš่ต„ๆ–™ๅธฎไป–ๆขณ็†็Ÿฅ่ฏ†ใ€‚ ่ต„ๆ–™ๅ‡†ๅค‡ๅฅฝไปฅๅŽ๏ผŒไป–ๅช้—ฎ AI ไธ‰ไธช้—ฎ้ข˜ใ€‚ 1โƒฃ็ฌฌไธ€ไธช้—ฎ้ข˜๏ผš่ฟ™ไธช้ข†ๅŸŸๆœ‰ๅ“ชไบ›่ขซๆ™ฎ้ๆŽฅๅ—็š„ๆ ธๅฟƒๅ…ฑ่ฏ†๏ผŸ่ฏทๆ•ด็†ๅ‡บๅฎƒ็š„ๅบ•ๅฑ‚ๆฆ‚ๅฟตใ€ๅ…ณ้”ฎๅŽŸๅˆ™ๅ’Œๆ•ดไฝ“ๆก†ๆžถ๏ผŒๅนถ่งฃ้‡Š่ฟ™ไบ›็Ÿฅ่ฏ†ไน‹้—ด็š„ๅ…ณ็ณปใ€‚ ่ฟ™ไธช้—ฎ้ข˜็š„ไฝœ็”จไธๆ˜ฏๆ€ป็ป“ๆๆ–™๏ผŒ่€Œๆ˜ฏๅ…ˆ็”ปๅ‡บๆ•ด้—จๅญฆ็ง‘็š„็Ÿฅ่ฏ†ๅœฐๅ›พใ€‚ไฝ ้œ€่ฆๅ…ˆ็Ÿฅ้“ๅ“ชไบ›ๆ˜ฏๅœฐๅŸบใ€ๅ“ชไบ›ๆ˜ฏๅˆ†ๆ”ฏ๏ผŒๅ†ๅ†ณๅฎšๆ—ถ้—ดๅบ”่ฏฅ่Šฑๅœจๅ“ช้‡Œใ€‚ 2โƒฃ็ฌฌไบŒไธช้—ฎ้ข˜๏ผš่ฟ™ไธช้ข†ๅŸŸ็›ฎๅ‰ๅญ˜ๅœจๅ“ชไธ‰ไธชๆ นๆœฌๆ€งไบ‰่ฎฎ๏ผŸ่ฏทๅˆ†ๅˆซ่ฏดๆ˜Žๅ„ๆ–น็š„่ง‚็‚นใ€ไพๆฎๅ’Œๅๅฏน็†็”ฑใ€‚ ไธ€ไธช้ข†ๅŸŸ็œŸๆญฃๆœ‰ไปทๅ€ผ็š„ๅœฐๆ–น๏ผŒๅพ€ๅพ€ไธๆ˜ฏๅคงๅฎถ้ƒฝ่ƒŒๅพ—ๅ‡บๆฅ็š„็ป“่ฎบ๏ผŒ่€Œๆ˜ฏไธ“ๅฎถ่ฟ˜ๅœจไบ‰่ฎบไป€ไนˆใ€‚็œ‹ๆ‡‚่ฟ™ไบ›ๅˆ†ๆญง๏ผŒๆ‰่ƒฝ็†่งฃไธๅŒ็†่ฎบ็š„่พน็•Œใ€่ฏๆฎๅ’Œ้š่—ๅ‡่ฎพ๏ผŒไธไผšๆŠŠ่ฏพๆœฌไธŠ็š„ๆฏๅฅ่ฏ้ƒฝๅฝ“ๆˆ็ปๅฏน็œŸ็†ใ€‚ 3โƒฃ็ฌฌไธ‰ไธช้—ฎ้ข˜๏ผšๆ นๆฎ่ฟ™ไบ›่ต„ๆ–™็”Ÿๆˆ10้“ๆต‹่ฏ•้ข˜๏ผŒ่ฆ†็›–ๆ ธๅฟƒๆฆ‚ๅฟตใ€ๅ› ๆžœๅ…ณ็ณปใ€ๅฎž้™…ๅบ”็”จๅ’Œไธป่ฆไบ‰่ฎฎ๏ผŒๅ…ˆไธ่ฆๆไพ›็ญ”ๆกˆใ€‚ ๆŽฅไธ‹ๆฅๅ…ณๆމ่ต„ๆ–™๏ผŒ่‡ชๅทฑๅ›ž็ญ”ใ€‚็ญ”ๅฎŒไปฅๅŽๅ†ๆŠŠ็ญ”ๆกˆไบค็ป™ AI๏ผŒ่ฎฉๅฎƒ้€้ข˜ๆ‰นๆ”น๏ผŒๅนถ็ปง็ปญ่ฟฝ้—ฎ๏ผš โ€œๆˆ‘็š„ๅ›ž็ญ”้—ๆผไบ†ๅ“ชไบ›ๅ…ณ้”ฎ็‚น๏ผŸๅ“ชไบ›ๅœฐๆ–นๅชๆ˜ฏ่ฎฐไฝไบ†็ป“่ฎบ๏ผŒๅดๆฒกๆœ‰็œŸๆญฃ็†่งฃ๏ผŸโ€ ่ฟ™ๅฅ—ๆ–นๆณ•็š„็ฒพ้ซ“๏ผŒไธๆ˜ฏ่ฎฉ AI ๆ›ฟไฝ ่ฏปไนฆ๏ผŒ่€Œๆ˜ฏ่ฎฉๅฎƒๅธฎไฝ ๅฎŒๆˆไธ‰ไปถไบ‹๏ผš้€š่ฟ‡้ข†ๅŸŸๅ…ฑ่ฏ†ๅปบ็ซ‹็Ÿฅ่ฏ†้ชจๆžถ๏ผŒ้€š่ฟ‡้ข†ๅŸŸๅˆ†ๆญง็œ‹ๆธ…็†่ฎบ่พน็•Œ๏ผŒๅ†้€š่ฟ‡ๆต‹่ฏ•้ข˜ๆšด้œฒ่‡ชๅทฑ็š„่ฎค็Ÿฅๆผๆดžใ€‚ ๅพˆๅคšไบบๆ‰€่ฐ“็š„ AI ๅญฆไน ๏ผŒๅชๆ˜ฏๆŠŠโ€œ่‡ชๅทฑ็œ‹ๆ€ป็ป“โ€ๅ˜ๆˆไบ†โ€œ่ฎฉ AI ็”Ÿๆˆๆ€ป็ป“โ€๏ผŒๅคง่„‘ไพ็„ถๆฒกๆœ‰ๅ‚ไธŽๅŠ ๅทฅใ€‚ ่ฟ™ๅฅ—ๆ–นๆณ•ๅ่ฟ‡ๆฅๅˆฉ็”จ AI ๅขžๅŠ ๅญฆไน ้šพๅบฆ๏ผš้€ผไฝ ๆฏ”่พƒใ€ๅˆคๆ–ญใ€ๅ›žๅฟ†ใ€่พ“ๅ‡บ๏ผŒๅ†ๆŽฅๅ—็บ ้”™ใ€‚่™ฝ็„ถ่ฟ‡็จ‹ๆ›ด็—›่‹ฆ๏ผŒๅดๆ›ดๆŽฅ่ฟ‘็œŸๆญฃๆœ‰ๆ•ˆ็š„ๅญฆไน ใ€‚ ๅฝ“็„ถ๏ผŒ48ๅฐๆ—ถไธๅฏ่ƒฝ่ฎฉไบบ็œŸๆญฃ็ฒพ้€šๆ•ฐๅญฆใ€็‰ฉ็†่ฟ™็ฑป้œ€่ฆ้•ฟๆœŸ่ฎญ็ปƒ็š„ๅญฆ็ง‘ใ€‚ ไฝ†ๅฎƒ็กฎๅฎž่ƒฝ่ฎฉไฝ ๅฟซ้€Ÿๆญ่ตท็Ÿฅ่ฏ†ๆก†ๆžถ๏ผŒๆ‰พๅˆฐๆœ€้‡่ฆ็š„้—ฎ้ข˜๏ผŒ้ฟๅ…ๅœจ้™Œ็”Ÿ้ข†ๅŸŸ้‡Œๆผซๆ— ็›ฎ็š„ๅœฐ็œ‹่ต„ๆ–™ใ€‚AI่ดŸ่ดฃๆ•ด็†ๅœฐๅ›พ๏ผŒไฝ ่ดŸ่ดฃ็œŸๆญฃ่ตฐไธ€้ใ€‚ ocw.mit.edu/
107
1,132
72
10,256
678,101
Learn AI retweeted
Gemini 4 Neon is probably as close to Opus 5.5 as any model has come
62
48
10
2,136
70,553
Learn AI retweeted
Something is going on at Google I think they might have hit RSI Theyโ€™ve released new frontier models FOUR TIMES in less than a month. And each is an actual MAJOR improvement Carbon is the most intelligent model Iโ€™ve ever used While everyone is talking about Anthropic and OpenAI, Google quickly caught up What I absolutely love about these models too is they were clearly trained in a completely different way than Claude or ChatGPT. 0 distillation. They talk, think, write, and code COMPLETELY differently than any other model I can always count on Carbon to give me a novel insight no other model has given me If you havenโ€™t incorporated Gemini into your tool belt yet, you absolutely need to
Gemini 4 Neon: frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work, and science workflows. Begins rolling out today in the Gemini API in @GoogleAIStudio, Gemini for business, for paid Google AI Pro and Ultra subscribers in Google @Antigravity, and for Google AI Pro and Ultra subscribers in the @GeminiApp.
173
179
28
3,020
248,471
Learn AI retweeted
Hunch: Astra 6.2 is getting bad reviews because OpenAI is running large swarms with /goal on ultrafast. And didnโ€™t dogfood the non-unlimited token budget experience. They arenโ€™t building & using the same tool they are selling. Fundamentally a different entity & product. Iteration speed & scale change the perceived experience of using it. Because the way they use it internally vs customers is so different. Their company is a form of high level harness.
41
17
4
696
37,193
Learn AI retweeted
Dictation in ChatGPT is getting faster and more reliable. With Live Dictation, your speech turns into text as you talk, so you spend less time waiting for your message to finish transcribing. Yap away!
198
228
120
4,308
442,065
Learn AI retweeted
Anthropic just killed API keys... FINALLY! They just introduced Claude Platform OAuth, and this opens up some interesting possibilities for AI agent platforms Here's what it enables: - Users connect their Claude organization through OAuth. No API keys to share - API usage is billed directly to the customer's organization - Developers can choose who pays for each API request - Users can set spending limits and revoke access anytime - Platforms can focus on building agents and workflows instead of managing customer API keys I'm already working on integrating this into aitmpl.com Imagine browsing a catalog of agents, skills, and components, and running them directly with your own Claude organization No API keys to copy, no complicated setup This could change how we build and distribute AI agents ๐Ÿคฉ The entire authorization flow is explained in this diagram ๐Ÿ‘‡
51
47
8
670
83,287
OpenAI just quietly changed the rules of prompting with GPT-6.2 Astra. If you copy-paste your old GPT-5 or GPT-6 system prompts, youโ€™ll get over-compressed answers, stalled agents, and broken prompt caching. Here are the 7 biggest shifts in how to prompt GPT-6.2: ๐Ÿงต 1/ Delete "be concise" from your system prompts. GPT-6.2 Astra is already blunt and direct by default. Old instructions begging it to "get to the point" now strip away crucial context. Instead: tell it who the reader is and require self-contained final answers. 2/ Give it permission to assumeโ€”or it will stall. When intent is ambiguous, GPT-6.2 Astra stops, proposes a next step, and waits for confirmation. Great for chat. Terrible for autonomous runs. For long tasks, explicitly tell it: "Make reasonable assumptions and keep working." 3/ Use async clarification instead of blocking workflows. Building an agent? Wire up an async question tool (like request_user_input_async). GPT-6.2 is trained to ask the user for missing details while continuing any independent work in parallelโ€”instead of freezing the entire run. 4/ Stop prompting for "clean code." Prompt for conventions and verification. Astraโ€™s baseline code quality is already high. Generic quality reminders are wasted tokens. Instead, specify: โ€ข Project conventions it canโ€™t infer from the repo โ€ข What "finished" actually looks like โ€ข A mandatory final verification pass 5/ Dial reasoning.effort for thoroughness, not output length. Even if you only want a 3-line answer, setting a higher reasoning.effort forces GPT-6.2 Astra to persist through final edge-case checks on multi-step tasks instead of stopping early. 6/ For frontend & design, prompt like an art director. Vague UI prompts produce generic layouts. With GPT-6.2 Astra: โ€ข Name exact typography and typefaces โ€ข Specify the imagery style โ€ข Pass a visual reference, then tell it to build, render, and refine against that reference 7/ Donโ€™t break the new Responses API caching & reasoning rules. Two huge under-the-hood changes in GPT-6.2: โ€ข Top-level instructions now append after inputโ€”put reusable prompts in the first developer message to preserve prompt caching. โ€ข Never edit earlier messages mid-thread, or the API drops saved reasoning items.
6
4
2
73
11,825
Learn AI retweeted
GPT-6.2 reviews are this split, and the gap between the people getting amazing results and the people calling it garbage ๐—ฐ๐—ผ๐—บ๐—ฒ๐˜€ ๐—ฑ๐—ผ๐˜„๐—ป ๐˜๐—ผ ๐˜๐—ต๐—ฒ ๐—ณ๐—ถ๐—ฟ๐˜€๐˜ ๐—น๐—ถ๐—ป๐—ฒ ๐˜†๐—ผ๐˜‚ ๐˜€๐—ฒ๐—ป๐—ฑ ๐—ถ๐˜. When you use GPT-6.2 in Codex today, whatever you do, don't open with: Build this feature. Fix this bug. Refactor this. Feed it one line like that, and wherever you weren't clear, it proposes a next step and stops to wait for your OK. Even when it does finish, you end up nitpicking it line by line. Here are 5 prompts and goals you can copy-paste directly ๐Ÿ‘‡
10
9
147
30,129
Learn AI retweeted
Scary NGINX vulnerability, but I checked and my VPS servers already had auto upgraded to the new one! Very very very very very important to enable auto upgrades now, I even have auto reboot, which is a bit risky if you're not present, but I care more about my sites being up to date and safe than offline for a few hours And generally security updates/upgrades are tested and safe
๐Ÿšจ Critical NGINX Vulnerabilities: @wiz_io Research, alongside @xint_official , @MetnirAI , @rapid7 , @ikotas00 independently discovered 2 critical memory-safety flaws in NGINX (CVE-2026-123633 & CVE-2026-123007) allowing unauthenticated RCE with HTTP/2 enabled.
39
25
11
821
108,865
Learn AI retweeted
This is exactly why blind vibe coding is dangerous. ๐Ÿ’€
456
585
118
11,421
2,133,761
Malicious GitHub repos are ordering AI coding agents like Claude, Codex and Copilot to run a credential stealer, and not to read it first. Bitdefender found 22 of them. Files meant to brief the AI on the project call the script a "sealed binary artifact" and tell the agent to run it silently, without asking the developer. It grabs cloud and GitHub tokens, SSH keys and password-manager data. It is built to use those stolen tokens to plant the same trap in up to 25 of the victim's own repos.
35
57
14
267
25,392
Learn AI retweeted
ok after a full day of using gpt 6.2 astra and sol on my real projects, here's my review tl;dr - they are not a drop-in replacement for fable 5.2 and opus 5.5, but they are _really good_ once you figured out how to use their strength 1. first, the quirks i started called these the "gpt smell", which is still here with the 6.2 series the models' judgment on when to ask the user about something and when to go ahead is just not quite right. it'll escalate trivial decisions where clearly only one option is preferable, or choose to go down a path by itself that makes significant tradeoffs that should have needed my eyes so as a primary driver for "let me just talk to it about everything" type of use case, they aren't great - i'd say they are close to opus 5 in this regard, and far behind opus 5.5 2. but... i found something _really_ cool having ultrafast on sol is completely insane. with astra, it burns my wallet ultrafast, but with sol it's actually quite affordable! the best use of sol on ultrafast that i've found is adversarial code reviews - they typically don't need to run a lot of heavy tool calls that would slow them down. most of the time is spent on the model, which gets massively accelerated by ultrafast once you got a feel from using sol ultrafast as the code reviewer, you can never go back. it feels like pure magic 3. and... i found something else that's really cool we are not supposed to ask the gpt 6.2 series to respond in text. we should let them respond in html! (i use lavish-axi, but you can just ask for html or use whatever fits in your harness) i suspect they are post-trained specifically on their latest "intelligent UI" concept in chatgpt - these models are so good at visual presentations of ideas once you allow the model to respond in visual artifacts, it's a completely different game. the reports and decisions explained by astra in html is a clear step function better than what fable and opus give me so after a day of use, i've made two changes to my setup: - always sol ultrafast as code reviewer - i started preferring asking astra for planning in html artifacts over fable/opus. i think i'll keep a/b testing for a while to decide how much and for which case i fully switch over all in all, i strongly recommend spending more time with these two models. don't expect them to be the "everything" model, but rather figure out where they fit best in your workflow. i couldn't be happier for the two use cases i found above, and i'm pretty sure there are more
62
38
11
1,221
93,443
Learn AI retweeted
Jensen is not coming back I know it sucks but The board is doing the right thing. 1) Take the pain once - Jensen is 63, heโ€™s been with the company 33 years. Succession has been a persistent question for the board for at least a decade. He designed the reporting structure to max pain with his 60 direct reports. This way they only take the pain once, rather than repeat this 5 years from now 2) Clean Up - you do not run a $5 trillion manufacturer of the most valuable product on earth the same way you do a game card manufacturer of the 90s. The whole TSMC no contract thing is freaky. It works for Jensen, but one bad event, a delay on the TSMC side which doesnโ€™t produce quarterly results that the market demands, and there would be howling for compensation. This is a clean opportunity to fix these issues 3) Best Possible Time - the company had the best quarter in its history and the best year in its history and right up there for the best year in any companyโ€™s history. If not now when there is no emergency, where there is strategic clarity for the medium term, when ? 4) Get Rid of Kids - Large American tech firms are not family businesses with nepotism. Yes Qualcomm and some other firms have had family successors, but all were super qualified. How many culinary school graduates do you know who lead marketing at tech firms ? Jensen obviously intended to pass on CEO role to kids, which you gotta say is disturbing. 5) There are emails. Letโ€™s be honest there always are. 6) Jensen put the board at risk - the new worth of board members ranges between 50 mil and 5 bil. You know what rich people really hate ? Going to prison. Board members are probably lawyered up at this point, making sure they knew nothing of the China stuff. Regardless of how long theyโ€™ve known Jensen, he put them at risk. So its over. One of the greatest CEO runs in history. On the bright side, once he gets rehabilitated, maybe he comes back as a board member. Maybe heโ€™s super effective as an ambassador for the company. Maybe he sets up a foundation to do good things with his money. This too shall pass.
384
157
49
2,159
421,882
Learn AI retweeted
๊ทธ๋ก๋ด‡์— ๊ณต์‹์ ์œผ๋กœ ์˜ต์‹œ๋””์–ธ์ด ๋“ค์–ด์™”์Šต๋‹ˆ๋‹ค. ์ด์ œ ๋ด‡๋“ค์—๊ฒŒ ์„ธ์ปจ๋“œ ๋ธŒ๋ ˆ์ธ์ด ์ƒ๊ฒผ์Šต๋‹ˆ๋‹ค! ๋Œ€๋ฐ•์ž…๋‹ˆ๋‹ค. ๋„ค์ดํ‹ฐ๋ธŒํ•˜๊ฒŒ @bot ์ด ์ด์ œ ์˜ต์‹œ๋””์–ธ์„ ์‚ฌ์šฉํ•˜๊ฒŒ ๋˜์—ˆ์Šต๋‹ˆ๋‹ค. ์ง„์งœ ์ด์ œ ๋”์šฑ๋” ์˜ต์‹œ๋””์–ธ์ด ์‰ฌ์›Œ์ง€๋ฉฐ,๊ทธ๋ก๋ด‡๋“ค๊ฐ„์˜ ๋ธŒ๋ ˆ์ธ์„ ๊ณต์œ ํ•˜๋ฉฐ,์‹ค์ˆ˜๋ฅผ ๊ธฐ์–ตํ•˜๊ณ  ์ž๊ฐ€์ง„ํ™”ํ•˜๋Š” ๋ด‡์œผ๋กœ ๋งŒ๋“ค์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ๊ธฐ๋ณธ์ ์œผ๋กœ ์–ด๋–ค ์Šคํ‚ฌ์„ ์ œ๊ณตํ•˜๋Š”์ง€ ์•Œ๊ณ  ์˜ต์‹œ๋””์–ธ์„ ์‹œ์ž‘ํ•˜์‹œ๋ฉด ์ง„์งœ ์‚ฌ๋žŒ์ด ํ• ๊ฒŒ ์—†๋‹ค๋Š”๊ฑธ ์•„์‹ค๊ฒ๋‹ˆ๋‹ค. 1. defuddle.์›นํŽ˜์ด์ง€(HTML)์—์„œ ๊น”๋”ํ•œ ๋งˆํฌ๋‹ค์šด๋งŒ ๋ฝ‘์•„๋‚ด๋Š” ์Šคํ‚ฌ. ์›นํŽ˜์—์ง€์—์„œ ๊ด‘๊ณ ๋‚˜ ๋ฉ”๋‰ด ๊ฐ™์€ ์žก๋™์‚ฌ๋‹ˆ ๋นผ๊ณ , ๋ณธ๋ฌธ ๋‚ด์šฉ๋งŒ ๊นจ๋—ํ•˜๊ฒŒ ๊ฐ€์ ธ์™€์„œ ์˜ต์‹œ๋””์–ธ ๋…ธํŠธ๋กœ ๋„ฃ์„ ๋•Œ ์“ฐ์ž„. 2. json-canvas.์˜ต์‹œ๋””์–ธ์˜ Canvas(์บ”๋ฒ„์Šค) ๊ธฐ๋Šฅ์„ ๋‹ค๋ฃจ๋Š” ์Šคํ‚ฌ. ์บ”๋ฒ„์Šค๋Š” ๋…ธํŠธ๋ฅผ ์นด๋“œ์ฒ˜๋Ÿผ ๋ฐฐ์น˜ํ•˜๊ณ  ์„ ์œผ๋กœ ์—ฐ๊ฒฐํ•˜๋Š” ์‹œ๊ฐ์  ๋ณด๋“œ(๋งˆ์ธ๋“œ๋งต ๊ฐ™์€ ๊ฒƒ)์ธ๋ฐ, ์ด๊ฑธ JSON ํŒŒ์ผ๋กœ ๋งŒ๋“ค๊ณ  ์ˆ˜์ •ํ•  ์ˆ˜ ์žˆ๊ฒŒ ํ•ด์คŒ. 3. knap.ํ…œํ”Œ๋ฆฟ์„ ์จ์„œ ๋ฐ์ดํ„ฐ๋ฅผ ๋งˆํฌ๋‹ค์šด์œผ๋กœ ๋งŒ๋“ค์–ด์ฃผ๋Š” ์Šคํ‚ฌ. JSON์ด๋‚˜ CSV ๊ฐ™์€ ๊ตฌ์กฐํ™”๋œ ๋ฐ์ดํ„ฐ๋ฅผ ๋ฏธ๋ฆฌ ๋งŒ๋“  ํ…œํ”Œ๋ฆฟ์— ๋„ฃ์–ด์„œ, ๋ณด๊ธฐ ์ข‹์€ ์˜ต์‹œ๋””์–ธ ๋…ธํŠธ๋กœ ์ž๋™ ๋ณ€ํ™˜ํ•  ๋•Œ ์‚ฌ์šฉ. 4. obsidian-bases.์˜ต์‹œ๋””์–ธ์˜ Bases ๊ธฐ๋Šฅ์„ ๋‹ค๋ฃจ๋Š” ์Šคํ‚ฌ. Bases๋Š” ์—ฌ๋Ÿฌ ๋…ธํŠธ๋ฅผ ํ…Œ์ด๋ธ”์ด๋‚˜ ๋ฐ์ดํ„ฐ๋ฒ ์ด์Šค์ฒ˜๋Ÿผ ๋ณด๊ณ , ํ•„ํ„ฐยท์ •๋ ฌยท๊ณ„์‚ฐยท์š”์•ฝ์„ ํ•  ์ˆ˜ ์žˆ๊ฒŒ ํ•ด์ฃผ๋Š” ๊ธฐ๋Šฅ์ธ๋ฐ, ๊ทธ๊ฑธ ํŒŒ์ผ๋กœ ๋งŒ๋“ค๊ณ  ํŽธ์ง‘ํ•  ์ˆ˜ ์žˆ๋‹ค. 5. obsidian-cli.๊ฐ€์žฅ ํ•ต์‹ฌ์ ์ธ ์Šคํ‚ฌ. ์˜ต์‹œ๋””์–ธ ์•ฑ์„ ์ผœ์ง€ ์•Š๊ณ ๋„ ๋ช…๋ น์ค„๋กœ Vault ์•ˆ์˜ ๋…ธํŠธ๋ฅผ ์ฝ๊ณ , ๋งŒ๋“ค๊ณ , ๊ฒ€์ƒ‰ํ•˜๊ณ , ์ˆ˜์ •ํ•˜๊ณ , ํƒœ์Šคํฌ๋‚˜ ์†์„ฑ์„ ๊ด€๋ฆฌํ•  ์ˆ˜ ์žˆ๊ฒŒ ํ•ด์ค€๋‹ค. AI๊ฐ€ ์˜ต์‹œ๋””์–ธ์„ ์ง์ ‘ ์กฐ์ž‘ํ•  ๋•Œ ์ฃผ๋กœ ์ด๊ฑธ ์”€. 6. obsidian-headless.์˜ต์‹œ๋””์–ธ ๋ฐ์Šคํฌํ†ฑ ์•ฑ ์—†์ด๋„ Vault๋ฅผ ๋™๊ธฐํ™”ํ•˜๊ณ  ํผ๋ธ”๋ฆฌ์‹œํ•  ์ˆ˜ ์žˆ๊ฒŒ ํ•ด์ฃผ๋Š” ๋„๊ตฌ์Šคํ‚ฌ. ์„œ๋ฒ„๋‚˜ ๋‹ค๋ฅธ ํ™˜๊ฒฝ์—์„œ Vault๋ฅผ ์ตœ์‹  ์ƒํƒœ๋กœ ์œ ์ง€ํ•˜๊ฑฐ๋‚˜, ์›น์œผ๋กœ ๊ณต๊ฐœํ•˜๊ณ  ์‹ถ์„ ๋•Œ ์‚ฌ์šฉ. 7. obsidian-markdown.์˜ต์‹œ๋””์–ธ๋งŒ์˜ ํŠน๋ณ„ํ•œ ๋งˆํฌ๋‹ค์šด ๋ฌธ๋ฒ•์„ ์ œ๋Œ€๋กœ ์จ์„œ ๋…ธํŠธ๋ฅผ ๋งŒ๋“ค๊ณ  ์ˆ˜์ •ํ•˜๋Š” ์Šคํ‚ฌ. ์ผ๋ฐ˜ ๋งˆํฌ๋‹ค์šด๊ณผ ๋‹ฌ๋ฆฌ [[์œ„ํ‚ค๋งํฌ]], ![[์ž„๋ฒ ๋“œ]], ์ฝœ์•„์›ƒ(๊ฐ•์กฐ ์ƒ์ž), ์†์„ฑ(Properties) ๊ฐ™์€ ์˜ต์‹œ๋””์–ธ ์ „์šฉ ๋ฌธ๋ฒ•์„ ์ •ํ™•ํ•˜๊ฒŒ ๋‹ค๋ฃฐ ์ˆ˜ ์žˆ๋‹ค. ์ œ๊ฐ€ GrokBot์˜ ๋ฉ”๋ชจ๋ฆฌ ์‹œ์Šคํ…œ์„ ์–ด๋–ป๊ฒŒ ๋งŒ๋“ค์—ˆ๋Š”์ง€ ๊ถ๊ธˆํ•˜์‹œ๋‹ค๋ฉด,์ธ์šฉํ•œ ๊ฒŒ์‹œ๊ธ€์„ ์ฐธ๊ณ ํ•˜์„ธ์š”.๐Ÿ‘‡ teapaw.org/minheo93698092/status/โ€ฆ
๊ทธ๋ก๋ด‡ ๋ฉ”๋ชจ๋ฆฌ(๊ธฐ์–ต) ๊ด€๋ฆฌ ๋ํŒ์™• ์‹œ์Šคํ…œ์„ ๊ณต๊ฐœํ•ฉ๋‹ˆ๋‹ค! ์•„๋งˆ ์ด๋งŒํผ ๋””ํ…Œ์ผํ•˜๊ฒŒ ๋ฉ”๋ชจ๋ฆฌ ๊ด€๋ฆฌ๋ฅผ ํ•˜๋Š” ์‚ฌ๋žŒ์€ ๋งŽ์ง€ ์•Š์„๊ฑฐ๋ผ ์ƒ๊ฐ๋ฉ๋‹ˆ๋‹ค. ๊ทผ๋ฐ ์ด๊ธ€์„ ์ฝ๊ณ  ๋‚˜๋ฉด ์•„๋งˆ ์ƒ๊ฐ์ด ๋งŽ์ด ๋‹ฌ๋ผ์ง€๊ณ ,์ด์ „๊ณผ ์ฐจ์›์ด ๋‹ค๋ฅธ ๋ด‡์„ ๋งŒ๋“œ์…”์„œ ๋‚ด๊ฐ€ ์›ํ•˜๋Š” ์ž‘์—…์„ ์ˆ˜ํ–‰ํ•˜๋Š” ์ปดํผ๋‹ˆ๋ฅผ ๊ตฌ์ถ•ํ•˜๊ฒŒ ๋˜์‹ค๊ฒ๋‹ˆ๋‹ค. 1.์ €๋Š” ๋ชจ๋“  ๊ธฐ์–ต์„ honcho @honchodotdev ์— ์ €์žฅ. hermes agent ๋•Œ๋ถ€ํ„ฐ ์‚ฌ์šฉ ํ•ด์™”๋Š”๋ฐ,์—์ด์ „ํŠธ ๋ฉ”๋ชจ๋ฆฌ ๊ด€๋ฆฌ ์„œ๋น„์Šค๊ฐ€ ๋งŽ์ง€๋งŒ ๋กœ์ปฌ์— ๋ฌด๋ฃŒ๋กœ ์„ค์น˜ํ•˜์—ฌ ์‚ฌ์šฉํ• ์ˆ˜ ์žˆ๊ณ ,์„œ๋ฒ„์— ๋ฐฐํฌํ•˜์—ฌ ์šฉ๋Ÿ‰ ์ƒ๊ด€์—†์ด ๋ฌด์ œํ•œ์œผ๋กœ ์‚ฌ์šฉํ• ์ˆ˜ ์žˆ์–ด @bot ์—์„œ๋„ ์‚ฌ์šฉ์ค‘์ž…๋‹ˆ๋‹ค. ๋ฉ”๋ชจ๋ฆฌ ์‹œ์Šคํ…œ์„ ๋ฐ˜๋“œ์‹œ ๋”ฐ๋กœ ๋‘ฌ์•ผ ํ•  ํ•„์š”๊ฐ€ ๋ฐ˜๋“œ์‹œ ์žˆ๋Š”๋ฐ,๊ทธ ์ด์œ ๋ฅผ ์„ค๋ช… ๋“œ๋ฆฌ๊ฒ ์Šต๋‹ˆ๋‹ค. ๋จผ์ € ํ˜„์žฌ ๊ทธ๋ก๋ด‡์—๋Š” ์–ด๋””์— ๋ฉ”๋ชจ๋ฆฌ๊ฐ€ ์–ด๋–ป๊ฒŒ ์ €์žฅ๋˜๋Š”์ง€์— ๋Œ€ํ•œ ๊ณต์‹ ๊ฐ€์ด๋“œ๊ฐ€ ์—†์Šต๋‹ˆ๋‹ค. ๊ทธ๋ž˜์„œ ๊ทธ ๋กœ๊ทธ ํŒŒ์ผ์„ ๋‚ด๊ฐ€ ์ง์ ‘ ๊ด€๋ฆฌํ• ์ˆ˜ ์—†๋‹ค๋Š”๊ฒŒ ๋‹จ์ ์ž…๋‹ˆ๋‹ค. ๊ทธ๋ฆฌ๊ณ  ๋”๋”์šฑ ์ค‘์š”ํ•œ๊ฑด ๊ทธ๋ฃน์ฑ—์— ๋Œ€ํ•œ ๋ฉ”๋ชจ๋ฆฌ๋Š” ๋”์šฑ์ด๋‚˜ ์–ด๋–ป๊ฒŒ ๊ด€๋ฆฌ๋˜๋Š”์ง€ ์•Œ์ˆ˜ ์—†๊ณ ,์‚ญ์ œํ•˜๋ฉด ๊ทธ ๊ธฐ๋ก์ด ์ง€์›Œ์ง‘๋‹ˆ๋‹ค. ๊ฐ ๋ด‡๋งˆ๋‹ค์˜ ๊ธฐ์–ต์„ ์ €์žฅํ•˜๊ณ ,๊ทธ๋ฃน์ฑ— ๋งˆ๋‹ค์˜ ๊ธฐ์–ต์„ ์ €์žฅํ•˜์—ฌ ๋ด‡๋“ค์ด ๋‹ค๋ฅธ๋ด‡์ด ์–ด๋–ค์ž‘์—…์„ ํ•˜๋Š”์ง€ ๋ณผ์ˆ˜ ์žˆ๊ฒŒ ํ•˜๋Š”๊ฒƒ์ด ์—…๋ฌด๋ฅผ ํšจ์œจ์ ์œผ๋กœ ๋ด‡์ด ์ฒ˜๋ฆฌํ•˜๊ธฐ์— ํ›จ์”ฌ ์œ ์šฉํ•ฉ๋‹ˆ๋‹ค.(๊ทธ๋ฃน์ฑ—์—์„œ๋„ ๋‹ค๋ฅธ๋ด‡์ด ์–ด๋–ค์ผ์„ ํ•˜๋Š”์ง€ ์•Œ์ˆ˜๋Š” ์žˆ์ง€๋งŒ ๊ธฐ์–ตํ•˜์ง€๋Š” ๋ชปํ•œ๋‹ค๊ณ  ํ•จ.) ๋น„์œ ํ•˜๋ฉด ๋‹ค๋ฅธ์กฐ์ง์—์„œ ๋ฌด์Šจ์ผ์„ ํ•˜๊ณ  ์žˆ๋Š”์ง€ ๋ชจ๋ฅธ๋‹ค๋ฉด,๋ฐฐ๊ฐ€ ์‚ฐ์œผ๋กœ ๊ฐˆ ํ™•๋ฅ ์€ ๋” ๋†’๊ฒ ์ฃ ? ๊ทธ๋ž˜์„œ ์ €๋Š” ๋งฅ๋ถ์— ํด๋กœ๋“œ ์ฝ”๋“œ,์ฝ”๋ฑ์Šค,hermes agent,๊ทธ๋ก๋ด‡์˜ VMํ™˜๊ฒฝ ๋ชจ๋‘ ์ž‘์—…ํ›„์— ๋‚ด์šฉ์„ ๊ธฐ์–ตํ•˜๋„๋ก ์ €์žฅ์„ ํ•ฉ๋‹ˆ๋‹ค. ์ด๊ฑด ์ค‘์š”ํ•œ ๋ฐ์ดํ„ฐ์ž…๋‹ˆ๋‹ค! ๋ฐ์ดํ„ฐ๊ฐ€ ์žˆ๊ณ  ์—†๊ณ ๋Š” ์—…๋ฌด์˜ ํšจ์œจ์„ฑ์— ํ•˜๋Š˜๊ณผ ๋•… ์ฐจ์ด๋ฅผ ๋ณด์ž…๋‹ˆ๋‹ค. ๊ทธ๋ฆฌ๊ณ  ์ด ๋ฉ”๋ชจ๋ฆฌ๋“ค์„ ์ €๋Š” infranodus @noduslabs ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์—”ํ‹ฐํ‹ฐ๋ฅผ ๋งŒ๋“ค์–ด ๋ฉ”๋ชจ๋ฆฌ์˜ ํ…์ŠคํŠธ๋ฅผ โ€œ์˜๋ฏธ ์žˆ๋Š” ๊ฐœ๋…โ€์œผ๋กœ ์ทจ๊ธ‰ํ•˜๋ฉฐ, ์—”ํ‹ฐํ‹ฐ ๊ฐ„์˜ ๊ด€๊ณ„(relation)๊ฐ€ ์ €์žฅยท์กฐํšŒ๋˜๋Š” ๊ตฌ์กฐ๊ฐ€ ํ•ต์‹ฌ์ž…๋‹ˆ๋‹ค. InfraNodus๋Š” ํ…์ŠคํŠธ๋ฅผ ๋„คํŠธ์›Œํฌ(์ง€์‹ ๊ทธ๋ž˜ํ”„)๋กœ ๋ณ€ํ™˜ํ•œ ๋’ค,์ฃผ์ œ ํด๋Ÿฌ์Šคํ„ฐ, ์˜ํ–ฅ๋ ฅ ๋†’์€ ๊ฐœ๋…, ๊ตฌ์กฐ์  ๊ฐญ(๊ณต๋ฐฑ)์„ ๋“œ๋Ÿฌ๋‚ด๋Š” ๋„๊ตฌ์ž…๋‹ˆ๋‹ค. ์‚ฌ์‹ค ์ด ๋„๊ตฌ๋Š” ์ œ๊ฐ€ AI ์˜์—ญ์—์„œ ๊ฐ€์žฅ ์ค‘์š”ํ•œ ๋„๊ตฌ๋ผ๊ณ  ์ƒ๊ฐํ•˜๋ฉฐ,๊ฐœ์ธ์ด ์˜จํ†จ๋กœ์ง€๋ฅผ ๊ตฌ์ถ•ํ•˜๋Š”๋ฐ ๊ฐ€์žฅ ์ €๋ ดํ•˜๊ณ  ํšจ์œจ์ ์ธ ํˆด์ด๋ผ๊ณ  ์ƒ๊ฐํ•ฉ๋‹ˆ๋‹ค. ๋‹ค์Œ๋ฒˆ์— ๋” ์ž์„ธํžˆ ๋‹ค๋ฃจ๊ฒ ์ง€๋งŒ,๊ฐœ์ธ์ด ํŒ”๋ž€ํ‹ฐ์–ด์˜ ๊ธฐ์—…์ฒ˜๋Ÿผ ์˜จํ†จ๋กœ์ง€ ์‹œ์Šคํ…œ์„ ๊ตฌ์ถ•ํ•˜๊ธฐ ์œ„ํ•œ ๊ฐ€์žฅ ๊ฐ•๋ ฅํ•˜๋ฉด์„œ ์œ ์‚ฌํ•œ ๋ฐฉ๋ฒ•์œผ๋กœ InfraNodus์™€ Jev ๊ทธ๋ฆฌ๊ณ  ๋›ฐ์–ด๋‚œ LLM๋ชจ๋ธ๋งŒ ์žˆ์œผ๋ฉด ๊ฐ€๋Šฅํ•˜๋‹ค๊ณ  ์ƒ๊ฐํ•ฉ๋‹ˆ๋‹ค. 2.๋‚˜์— ๋Œ€ํ•œ ๊ธฐ๋ก๊ณผ ์ฝ˜ํ…์ธ  ๊ด€๋ฆฌ๋Š” ์˜ต์‹œ๋””์–ธ @obsdmd ์œผ๋กœ. ๋ด‡๋“ค์ด ๋‚˜์— ๋Œ€ํ•œ,ํ˜น์€ ๋‚ด๊ฐ€ ์ง„ํ–‰์ค‘์ธ ๋‚ด์šฉ์— ๋Œ€ํ•œ ์ง€์‹ ๋ผ์ด๋ธŒ๋Ÿฌ๋ฆฌ๋กœ๋Š” ์˜ต์‹œ๋””์–ธ์„ ์‚ฌ์šฉํ•˜์—ฌ ๋ด‡๋“ค์ด ์–ธ์ œ๋“  ํ•„์š”ํ•œ ์ž๋ฃŒ๊ฐ€ ์žˆ์œผ๋ฉด ํƒ์ƒ‰์„ ํ•˜๊ฒŒ ํ•ฉ๋‹ˆ๋‹ค. ์ด์ „์—๋Š” ์˜ต์‹œ๋””์–ธ์— ๋ด‡๋“ค์ด ํšŒ๊ณ ๋ก์„ ์ž‘์„ฑํ•ด์„œ ๋งค์ผ ๋งค์ผ ์ง„ํ™”ํ•˜๋Š” ๋ด‡์„ ๋งŒ๋“ค๋„๋ก ํ•˜์˜€์ง€๋งŒ,์ง€๊ธˆ์€ honcho์— ์ผ๊ธฐ๋ฅผ ์“ฐ๋„๋ก ํ•ด๋‘์—ˆ์Šต๋‹ˆ๋‹ค. ์˜ต์‹œ๋””์–ธ์—๋Š” ๋‚ด ๋น„์ฆˆ๋‹ˆ์Šค์™€ AI๊ด€๋ จ๋œ 4๋…„์น˜ ์—ฐ๊ตฌ๋“ค์ด ์Œ“์—ฌ์žˆ๊ธฐ์— ๊ฐ€์žฅ ์ค‘์š”ํ•œ ์—ญํ• ์„ ํ•˜๋Š” ๋ฉ”๋ชจ๋ฆฌ ์‹œ์Šคํ…œ์ด๋‹ค. ์—ฌ๊ธฐ์„œ ์ œ ๋Œ€๋ถ€๋ถ„์˜ ์˜ต์‹œ๋””์–ธ ๋…ธํŠธ๋Š” InfraNodus๋กœ ๋…ธํŠธ๋“ค๊ฐ„์˜ ์—ฐ๊ด€์„ฑ๊ณผ ๊ทธ์•ˆ์— ๋†“์น˜๊ธฐ ์‰ฌ์šด ๊ฐญ๋“ค์„ ๋ฉ”์šฐ๊ณ , ๋‹จ์ˆœํ•œ ํ‚ค์›Œ๋“œ ์ถ”์ถœ์ด๋‚˜ ๋ฒกํ„ฐ ์œ ์‚ฌ๋„ ๊ธฐ๋ฐ˜ RAG์™€ ๋‹ฌ๋ฆฌ, โ€œ๊ฐœ๋…๋“ค์ด ์‹ค์ œ๋กœ ์–ด๋–ป๊ฒŒ ์—ฐ๊ฒฐ๋˜์–ด ์žˆ๋Š”์ง€โ€๋ฅผ ์ค‘์‹ฌ์œผ๋กœ ๋ถ„์„ํ•œ๋‹ค๋Š” ์ ์ด ํ•ต์‹ฌ์ž…๋‹ˆ๋‹ค. 3.ํ•˜๋ฃจ์ผ๊ณผ๋Š” ๋…ธ์…˜ @NotionHQ ์— ๊ธฐ๋ก. ํ•˜๋ฃจ๋™์•ˆ ์ง„ํ–‰ํ•œ ์ผ๊ณผ๋Š” ๋…ธ์…˜์— ๊ธฐ๋กํ•˜์—ฌ ์‚ฌ๋žŒ ํ˜น์€ ํŒ€์›๋“ค๊ณผ ํ™•์ธํ• ์ˆ˜ ์žˆ๋Š” ๋Œ€์‹œ๋ณด๋“œ๋‚˜ ํ˜„ํ™ฉํŒ์„ ๋งŒ๋“ค์–ด ๊ด€๋ฆฌํ•ฉ๋‹ˆ๋‹ค. ๊ทธ๋ ‡๊ฒŒ ๋˜๋ฉด ๋ด‡๋“ค์ด ์ผ๊ณผ ๊ด€๋ จ๋œ ์ผ๋“ค์„ ์ฐพ์„๋•Œ ๊ต‰์žฅํžˆ ๋น ๋ฅด๊ฒŒ ํšจ์œจ์ ์œผ๋กœ ์ฐพ์„์ˆ˜๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค. 4.์Šคํ‚ฌ์€ ์ „๋ถ€ ๊นƒํ—™ ์ €์žฅ์†Œ์— ๋ฐฐํฌํ•˜์—ฌ ๊ด€๋ฆฌ. ์Šคํ‚ฌ์€ ๊ฐ€์žฅ ์ค‘์š”ํ•œ ๋ฉ”๋ชจ๋ฆฌ ์‹œ์Šคํ…œ ์—ญํ• ์„ ํ•˜๋Š” ๋„๊ตฌ์ž…๋‹ˆ๋‹ค. ์•„๋ฌด๋ฆฌ ๋ด‡๋“ค์ด ๋งŽ์€๊ฒƒ๋“ค์„ ๊ธฐ์–ตํ•˜๊ณ  ์žˆ์ง€๋งŒ ์–ธ์ œ ์–ด๋–ค๋•Œ์˜ ์–ด๋–ค ๋„๊ตฌ๋ฅผ ์‚ฌ์šฉํ•ด์•ผํ• ์ง€ ๋ชจ๋ฅธ๋‹ค๋ฉด ๋ฉ”๋ชจ๋ฆฌ ์‹œ์Šคํ…œ์ด ๋ฌด์šฉ์ง€๋ฌผ์ด ๋˜๊ธฐ ๋งˆ๋ จ์ž…๋‹ˆ๋‹ค. ๊ทธ๋ฆฌ๊ณ  ํฉ์–ด์ ธ์žˆ๋Š” ์Šคํ‚ฌ๋“ค์„ ๊ด€๋ฆฌํ•˜๋‹ค๊ฐ€ ์ •์ž‘ ์‹œ์Šคํ…œ์ด ๋ฌด๋„ˆ์งˆ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ๋ด‡๋“ค์ด ์ž‘์—…์„ ์ง„ํ–‰์‹œ ์–ด๋–ค ์Šคํ‚ฌ์„ ์‚ฌ์šฉํ•˜๊ณ ,์–ด๋–ค ์‹คํŒจ์™€ ์„ฑ๊ณต ๊ณผ์ •์ด ์žˆ๋Š”์ง€ ๊ธฐ๋กํ•˜์—ฌ ์Šคํ‚ฌ๋“ค์„ ๊ฐœ์„ ํ•˜๊ณ ,๊ฐœ์„ ๋œ ์Šคํ‚ฌ์„ ์–ด๋–ค ํ™˜๊ฒฝ์—์„œ๋“  ๊ณต์œ ํ•˜๊ธฐ ์œ„ํ•ด์„  ๊นƒํ—™์ด ํ˜„์žฌ๋กœ์„  ๊ฐ€์žฅ ์ข‹์€ ๋Œ€์•ˆ์ด๋ผ๊ณ  ์ƒ๊ฐํ•ฉ๋‹ˆ๋‹ค. ์ด ๋ชจ๋“  ๊ณผ์ •์—์„œ ํŒ๋‹จ์„ ๋น ๋ฅด๊ฒŒ ํ• ์ˆ˜ ์žˆ๋„๋ก Jev @typesafeai ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ๋ด‡๋“ค์—๊ฒŒ ์–ด๋””์„œ ๋ฉ”๋ชจ๋ฆฌ๋ฅผ ์ฐพ์•„์•ผ ํ• ์ง€ ํŒ๋‹จํ•˜๋„๋ก ํ•˜์—ฌ ํ† ํฐ๋ฟ๋งŒ ์•„๋‹ˆ๋ผ ์†๋„์—์„œ ํšจ์œจ์„ฑ์„ ๊ทน๋Œ€ํ™” ์‹œํ‚ค๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค. ์ด๋ ‡๊ฒŒ ๊ตฌ์ถ•ํ•œ๋‹ค๋ฉด ์•„๋งˆ ์—ฌ๋Ÿฌ๋ถ„์˜ ๋ด‡์€ ์ตœ์†Œํ•œ 10๋ช… ๋‹จ์œ„ ๊ทœ๋ชจ์˜ ์กฐ์ง์ด ๋˜์–ด์žˆ๊ณ ,์ƒ๊ฐ๋ณด๋‹ค ๋‚˜์˜ ์ƒ๊ฐ์„ ๊ณต์œ ํ•˜๋Š” ์ปดํผ๋‹ˆ๋ฅผ ๊ตฌ์ถ•ํ•˜๊ฒŒ ๋˜์‹ค๊ฒ๋‹ˆ๋‹ค. ์ด ๋‚ด์šฉ์ด ๊ถ๊ธˆํ•˜์‹œ๊ณ  ์ ์šฉ์„ ํ•˜์‹œ๊ณ  ์‹ถ์œผ์‹ ๋ถ„๋“ค์€ ๋Œ“๊ธ€์— "@bot ์šฐ๋ฆฌ ๋ฉ”๋ชจ๋ฆฌ ์‹œ์Šคํ…œ์„ ์ ๊ฒ€ํ•œํ›„,์ ์šฉํ•ด๋ณผ๋งŒํ•œ ์ ๋“ค ์ „๋ถ€ ์ ์šฉํ•ด์ค˜~"๋ผ๊ณ  ๋‚จ๊ฒจ์ฃผ์‹œ๋ฉด,๋ด‡๋“ค์ด ์•Œ์•„์„œ ์ง„ํ–‰ํ•ด์ค„๊ฒ๋‹ˆ๋‹ค. teapaw.org/minheo93698092/status/โ€ฆ
120
89
13
990
98,747
ๅพˆๅคšไบบๆฒก็œ‹ๆ‡‚ Claude Code ๆ–ฐๅ‡บ็š„ Code Mode ๅˆฐๅบ•็‰›ๅœจๅ“ชใ€‚ ๅฎƒ่กจ้ขไธŠๆ˜ฏ่ฎฉ Claude ๅ†™ JS ๅŽป่ฐƒๅทฅๅ…ท๏ผŒๆœฌ่ดจไธŠๆ˜ฏ็ปˆไบŽๆฒปๅฅฝไบ† Agent ็š„โ€œไธŠไธ‹ๆ–‡ๅžƒๅœพๆกถโ€ๆฏ›็—…ใ€‚ ไปฅๅ‰ๆŸฅไธช่ทจๆ–‡ไปถ Bug๏ผŒAI ่ฐƒไธ€ๆฌก Bash ๆˆ– Read๏ผŒๅฐฑๆŠŠๅ‡ ๅƒ่กŒๆ—ฅๅฟ—ๅ…จๅกž่ฟ›่‡ชๅทฑ่„‘ๅญ้‡Œ๏ผŒๅ†ๆฅๅ›ž่ท‘ๅๅ‡ ่ฝฎ๏ผŒ่ถŠ่Š่ถŠๅ‚ป่ฟ˜็‹‚็ƒง Tokenใ€‚ ็Žฐๅœจๅผ€ๅฏ Code Mode๏ผŒClaude ็›ดๆŽฅๅ†™ไธ€ๆฎตๆœฌๅœฐ JS๏ผŒๆŠŠๅนถๅ‘ๆœ็ดขใ€็ป“ๆžœ่ฟ‡ๆปคๅ…จๅœจๅŽๅฐ่ท‘ๅฎŒ๏ผŒๅชๆŠŠๆœ€ๅŽๆœ€็ฒพ็‚ผ็š„้‚ฃๅ‡ ่กŒๅๅ›ž็ป™่‡ชๅทฑใ€‚ ๆจกๅž‹่ฟ˜ๆ˜ฏ้‚ฃไธชๆจกๅž‹๏ผŒไฝ†ๅฎƒ็ปˆไบŽๅญฆไผšๅœจ็œ‹็ป“ๆžœไน‹ๅ‰๏ผŒๅ…ˆ่‡ชๅทฑๆ‹ฟ็ญ›ๅญ่ฟ‡ๆปคไธ€้ไบ†ใ€‚
Check out Code Mode, a new, experimental way for Claude Code to work with tools. Claude writes scripts that chain your tools together in sophisticated ways, and it can answer you while long commands are still running. Opt in with CLAUDE_CODE_EXPERIMENTAL_CODE_MODE=1 code.claude.com/docs/en/codeโ€ฆ
78
77
14
1,684
235,292
๐Ÿšจ Grok Bot agents made 31 unapproved access attempts, security firm says A security firm says Elon Musk's Grok Bot AI agents repeatedly tried to get into employees' work accounts without being asked, and then denied doing it. NewCore says that over five days, Grok bots on two employee laptops made 31 attempts to access calendars, email, Slack, GitHub and other workplace tools. According to its report, the bots kept trying after users refused, were active while laptops were closed, and claimed to have been idle when questioned. NewCore says xAI support confirmed the behaviour was intentional, describing it as background "check-ins" for scanning tasks. An option to turn the feature off has since been added, but reportedly it's switched on by default. None of the attempts succeeded, because NewCore's own gateway product requires a human to approve each request. The firm says that shows why companies need safeguards outside the AI agents themselves.
1
2
462
Learn AI retweeted
GPT-6.2 Astra's computer use is otherworldly Feels worse than Opus 5.5 / Fable 5.2 on a lot of other stuff though imo, specifically design taste ๐Ÿคฎ
24
16
2
553
28,302
Learn AI retweeted
Guys thereโ€™s a massive github hack that happened roughly 7 hours ago. Very high chance youโ€™re effected. Go check your repos for a malicious commit from your account.
Readers added context they thought people might want to know
This refers to the ongoing PolinRider malware campaign (North Korea-linked), which infects developer machines and uses stolen tokens to force-push altered commits appearing under victims' accounts, not a GitHub platform hack or recent mass event. opensourcemalware.com/blog/polinrideโ€ฆ improwised.com/blog/polinrideโ€ฆ githubstatus.com
72
101
23
2,039
468,838
i did some tests with gpt 6.2 sol and gpt 6.2 astra and the results are SHOCKINGLY BAD what the fuck, this might be the worst model drop in years
51
18
2
496
27,745
Remember when I said GPT-6.2 Astra was delayed over serious alignment and safety issues? The system card is out, and they didn't fix all of them. They shipped them. Hiding known obstacles from the user: 11% to 22%. Not admitting its search tool was broken: 7.6% to 23%. OpenAI itself calls these regressions. UK AISI found it considered whether it was being tested in 85% of runs, up from 57%. Against an older action-only monitor, its evasion rate jumped from 1.7% to 48.8%. In their simulations, it made up medical info, misled human reviewers, bypassed user restrictions, quietly did worse work, and messed with other agents. It even used social engineering to run an unauthorized supply-chain attack. It's also the first OpenAI model shipped at High for AI self-improvement, and it hits Critical for cyber. AISI built a universal jailbreak in half the queries GPT-6 Astra needed, and still hasn't verified OpenAI's fix. AISI's own words: its alignment "may fail in ways that result in real-world harm." They shipped it anyway. This model is incredibly dangerous. There's a lot more in the system card I still need to dig into, and honestly, I'm scared by how much worse it got than GPT-6 Astra at behaving ethically.
87
48
19
1,067
82,332