ファンサなんか期待するより 推しが今もそこにいてくれてる 笑ってくれてる それだけで、十分なんよ。

@mossacannibali3

fara goons

이게바로 요즘 인기 트윗인... 인기순으로 보기입니까

[Feature] Why Japan’s Industry is Pulling Back from the U.S. Navy Frigate Plan 🇯🇵 🇺🇸 🏗️ Japanese defense industry stakeholders are growing hesitant over a potential U.S. Navy frigate deal, weighed down by heavy investment demands and domestic naval commitments. By @japanesepatrio6

AIで完全放置させているサイトのSEO改善プロンプト スキル名: SEO Rank Watch 対象サイトのGoogle検索順位を継続的に改善する。 目的：1位を取れそうなキーワードを見つけ、検索ニーズに答える改善を1つ行い、7日間観察する。これを1位になるまで繰り返す。 データ data/seo/watchwords.json — キーワード / 対象ページ / 優先度 data/seo/rank-history.json — 順位履歴。追記専用 data/seo/improvement-log.json — 改善履歴 / status / 次回レビュー日 status： active — 改善候補 observing — 改善後7日間の観察中 achieved — 1位達成。監視のみ Workflow 1. 順位を測る 原則としてGoogle Search Consoleを使用する。 node.claude/skills/seo-rank-watch/scripts/fetch_gsc_ranks.mjs \ --repo <REPO_PATH> --append GSCの平均…

iPhone Duo green screen

🚨只改一个文件，能让GPT-6 Astra Token直接少烧35%+！ GPT-6 Astra不是贵，是你的AGENTS.md太肥了 省Token只做一件事：把仓库根目录那份 AGENTS.md 换成精简版（如图） 修改后实测数据很狠： 🔹普通改代码省 5%–15% 🔹跨文件长任务省 10%–25% 🔹爱反复搜索、重复验证的任务能到 30%–35%+ GPT-6 Astra 协作规则真正砍Token的不是文字变短，是把Agent的坏习惯掐死： 🔸先给结论，不复读需求 🔸已确认的信息不重读、不重搜 🔸证据够了就停，找到原因就修 🔸已授权操作不反复确认 🔸验证力度跟这次风险对齐，不为“看起来严谨”加测试 🔸交付物出来、必要验证过、限制说清，立刻收工 🔸子Agent只在独立且收益大于协调成本时才拆 GPT-6 Astra对 AGENTS.md 更敏感。文件越全，它越容易被旧规则带着空转。 根文件只留稳定规则，细节按需加载，账单才会掉。 收藏对照改。改完拿你最费Token的那个长任务对比一次，很多时候省下的不是35%，是整段无效探索。 #GPT6 #Astra #AG…

【話題】“話し合い”ができない夫を激詰めした結果、離婚を切り出された女性が話題に ・夫と議論しようとすると「明日にして」と逃げられる ・夫は“話し合いができない人”で、大体何でも夫が悪いというスタンス →既婚男性から「読んでてツラい」「うちの嫁が書いた文章かと思った…」との感想が相次ぐ

VM？

Wait for Grandma 🤯 if you are married, I advise you not to open the commentsss

個人的に無料版CatGPT,Claudeでも一発で読み込めるadhx.com使ってるな〜 𝕏.comの先端に三文字追加するだけでいいし… (というかfxtwitter何故か読み込めなかった)

The life of the goyim.

Hand movement is a bit rusty but the spitting is on point 👍

Me explaining AI to friends

What’s the point of 4K? Honestly, most people use 4K for those hyper-detailed, slow-mo macro shots. And I get it—they work because they grab attention. But I’m interested in something else: a raw, realistic aesthetic that feels less like a digital file and more like a real film—the kind that forces you to stop and watch, rather than just scroll past. When @capcutapp invited me to try Seedance 2.0 4K, I didn't want to make just another promo ad. I wanted to see if I could push the tech into territory that feels like a real film. I set up a few stress tests to see if this 4K could handle cinematic storytelling: - Dark Environments: Low-light gradients that usually turn to digital mush at lower resolutions. - Vibrant Color: Hyper-saturated schemes that often band and block. - Macro Shots: Using those "attention-grabbing" angles, but giving them a narrative purpose rather than just using them as a gimmick. I channeled the gritty, flashing-light aesthetic of Gaspar Noé for this. (A quick warning: his films are intense—careful.) The result is "Salty." Seedance 2.0 4K handled the mood and the detail better than I expected. Whether you're making films for fun or for a living, it’s a powerful tool for bringing those wilder, cinematic ideas to life. Huge thanks to the team at CapCut for trusting me with early access to Seedance 2.0 4K. Getting to experiment within CapCut Video Studio was a true honor—I really appreciate the creative freedom they gave me for this film. Salty A film by Jordan Daniel Chesney Produced by CapCut Music by Brendon Moeller @capcutapp #capcut 4K YouTube version in the comments.

Introducing shadscan 🪓 Deterministic UI audits for React shadcn apps. No AI, no API keys. Fully open source. One command. No install. npx [@]shadscan/cli Built for your terminal and your CI. 🧵

@thsottiaux tl;dr version: - set up CLIProxyAPI with Claude and Codex auth - Connect to Claude Code - Make "claudex" alias that sets some env vars Took like 2 prompts (I already had the proxy set up tbf)

Launching our new paper on arXiv: we trained the largest multilingual food model ever built. 4.1M recipes. 7 languages. 1,790 ingredients. 300 dimensions. All of human cooking compressed into 2 megabytes.

Inspired by, but for football. Draft a Premier League XI from random clubs and seasons. Build your team. Simulate the season. Can you go 38-0? #football #soccer #fifa #premierleague

all the most successful startups are all doing this rn they have multiple accounts where they all post different formats, different content, and post them consistently push out 20 to 30 videos a day, if you post 300 to 1,000 videos a month there is no way that you are not going to succeed… but doing that all by yourself is basically trying to push a 50-pound boulder up a mountain the short cut loop to that is Content Rewards on Content Rewards, you can launch a campaign, put a budget suddenly overnight, you have 30+ accounts with your app's branding that are consistently posting for your brand because they want to earn the campaign money you can specify a viral format, and put them in a drive and get them to your creators creators will multiply the viral format and post 20, 30, even 50 times a day and all of this is before you even spend a dollar you only pay clippers/creators when they generate views for you $0.12 cents to $1 cpm on verified views only that's exactly how we're getting brands super bowl level views in such short time and spend

I want to master GPT images 2.0.

Claude Design is insane. ❤️‍🔥Just recorded a 18-min tutorial on how to build animated, award-winning websites with Claude Design + Opus 4.7!

Happy Halloween! 🎃 Voice: @saruei Voice/Sfx: @MimiHung_VO

For full transparency, here’s my track record over the last 6 months: I publicly called the exact Bitcoin top at $126K. I shorted ZEC at $717 (which ended up being the exact top again). I told you to buy OIH and XLE, both are up 12%+ in just a few days, but this is a multi-year trade. I bought NTR a few days ago (currently at breakeven). I’m still accumulating, but this is a three-year trade anyway. I don’t make a lot of calls because, in my opinion, good setups are usually rare. But when I make a move, you at least know my conviction is through the roof. My goal here is to become the first person on this app to reach 1M followers without making $1 from my audience. all while delivering as much knowledge as I possibly can. $10k, $100k, $1M, I don’t care. I’m not interested in promoting stuff. I don’t need money. Thank you. My tweets are very time-sensitive (the market moves fast). Turn on notifications so you don’t miss anything. Big things are coming this year. Let’s keep fighting Wall Street and win together 🤜🤛

So many people hating on how many lines of code this is and how it's "slop" But Claude Code still has the best memory, context recall and speed that all agent builders should learn from. 1. Loads memory from 6 priority layers, bottom to top: Org policy → user prefs → project rules → local overrides → auto-extracted memories → team-synced shared memory. Rules can be conditional: they only activate when you're touching matching files. Auto-recall uses a side-query to rank and pick the top 5 most relevant memories from 200+ candidates. The whole thing is memoized per session. It's closer to a filesystem than a prompt. 2. Compresses Context with 3 tiers: Tier 1: Replay a pre-extracted session summary ((min 10K tokens, max 40K). No API call. Fastest. Tier 2: Surgically prune old tool results from the cached prefix without rebuilding the cache. Cache-safe. Tier 3: Full conversation summarization via Sonnet. Last resort. Has a retry loop that progressively truncates the head if the compaction request itself is too long 3. Parallelization everywhere Startup fires keychain reads, git ops, and subprocess work all concurrently: I/O finishes during module import, effectively free (65ms). Nothing blocks first render. File search returns results in ~5ms before the index finishes building. Fork agents share byte-identical request prefixes for max prompt cache reuse. Tools declare themselves as concurrency-safe and execute in parallel during streaming. If one errors, siblings abort but the parent query continues. A function literally named DANGEROUS_uncachedSystemPromptSection() exists to scare devs away from breaking the cache. Nothing is allowed to be slow.

WELCOME TO THE SKILL ERA OF THE INTERNET for the last 15 years, if you wanted to build a serious software company, you built a product and exposed an api. that was the move. you created functionality… payments, messaging, email, search, analytics… and then you let developers plug into it. the companies that won owned the pipes. stripe owned payments. twilio owned messaging. sendgrid owned email. the api was the distribution layer. once you were integrated, you were embedded. that model made sense in a world where execution was scarce. llms compress execution into a prompt. so the center of gravity shifts. in this cycle, you build expertise and package it as a skill. an api is a doorway into a function. here’s how to send an email. here’s how to process a payment. here’s how to fetch this data. it’s precise. mechanical. bounded. a skill is a doorway into judgment. here’s how to audit a landing page like a serious growth operator. here’s how to structure a legal intake so you catch the real risk. here’s how to clean and enrich messy directory data so it actually turns into revenue. you’re encoding a way of thinking. and that changes how companies are built and how they scale. in the api era, distribution meant convincing developers to integrate you. you needed docs. sdk’s. developer evangelism. you fought for a place inside someone else’s codebase. in the skill era, distribution means becoming part of someone’s agent workflow. a founder opens claude code. they type /seo-audit. your skill runs. it frames the output. it structures the analysis. it guides the decisions. your expertise lives inside the execution layer itself. you aren’t pulling users into your interface. you’re embedding your thinking into theirs. that changes company design. the old playbook looked like this: build saas design ui onboard users drive retention expand seats the new playbook looks more like this: encode a high-leverage playbook package it as a skill let agents c…

Here’s how to slash sugar spikes from pasta — and make it healthier. Scientists discover that cooling cooked pasta for 24 hours transforms its starch into a gut-friendly fiber that slashes blood sugar spikes. When you cook pasta and then cool it for at least 24 hours, the starch molecules undergo a process called retrogradation. This molecular reorganization transforms digestible starch into "resistant starch," which functions more like fiber than a simple carbohydrate. Instead of being rapidly absorbed in the small intestine and causing a sharp rise in glucose, it travels to the large intestine where it feeds beneficial gut bacteria as a powerful prebiotic. This simple kitchen hack not only supports a healthy microbiome but also encourages the production of short-chain fatty acids that help reduce internal inflammation. Beyond gut health, this transformation provides a significant advantage for metabolic wellness by slowing sugar absorption and improving insulin response. Perhaps most surprisingly, these health benefits remain largely intact even if you choose to reheat the pasta before eating. By simply planning ahead and refrigerating your noodles overnight, you can effectively lower the caloric impact of your meal while maintaining more stable energy levels throughout the day. This game-changing strategy allows you to enjoy favorite comfort foods while prioritizing long-term health and better blood sugar control. Source: Bullen, A. (2022). Is Reheated Pasta Actually Better for You? Cleveland Clinic Health Essentials.

How to (actually) write content with AI {prompts included}