https://t.co/hsvsflUPmf

We ditched code review at @DuckbillHQ (mostly) About a month ago, we found ourselves with 60 open PRs for a team of five. They had been accumulating for a few weeks and we all had the sudden realization we were looking at two days of just code review.

70 Ideas to Reimagine Your Product Process With AI

so astra writes python to do everything which means we're going to have to rework this ui https://t.co/5cAttUtS6w

Here is how to enable a 1M-token context window in Codex for GPT-5.6 Sol. Even though we have tuned the context limit in Codex to be set optimally when it comes to performance and cost, this is a common ask, so here it is documented. A larger context window lets Codex retain more code, tool output, and conversation history before summarizing older material. You need a model that supports it. And GPT-5.6 Sol, for example, has a documented 1,050,000-token window. Open ~/.codex/config.toml and add or update these settings at the top level, before any [section] headers: ``` model = "gpt-5.6-sol" model_context_window = 1000000 model_auto_compact_token_limit = 900000 ``` The first setting selects the model. The second tells Codex to use a one-million-token context budget. The third starts automatic history compaction around 900,000 tokens, leaving some headroom. Restart Codex client and start a new session after saving. To try the configuration for a single CLI session without changing your defaults: ``` codex -m gpt-5.6-sol \ -c model_context_window=1000000 \ -c model_auto_compact_token_limit=900000 ``` Have fun, but also know that we tuned the default carefully!

If you’ve stopped doing human code reviews: how is it going? If you’re still doing them: how are you able to keep up with the explosion of code + PRs, and how do you prioritise what to review “properly”?

Best Open-source AI Models for Cybersecurity that run fully offline on low hardware & The comparison

Just ask Astra to remove all the unnecessary text. Everywhere.

China published the most uncomfortable paper on vibe coding. ETH Zurich tested 100 developers in a controlled, commercial-grade vibe coding environment to see who actually succeeds. The findings are brutal. The researchers tracked computer science achievement, written communication skills, and general cognitive reasoning. They wanted to see what actually predicts vibe coding proficiency when you never touch a line of source code yourself. Two major predictors emerged. Written communication proficiency mattered. The ability to structure thoughts and articulate intent unambiguously in text directly impacts what the AI builds. But that wasn't even the main takeaway. Computer science achievement was a massive, dominant predictor of success. Even when researchers controlled for general intelligence and reasoning skills, CS background still heavily dictated who built working software and who completely crashed. In fact, CS knowledge contributed roughly twice the unique predictive variance of writing skills alone. Why? Because vibe coding isn't about writing code. It’s about debugging logic. When an AI agent builds a complex application and quietly breaks an edge case under the hood, a non-technical user looks at the glowing UI and assumes it works. They don't know what questions to ask. They don't know what logic to challenge. They lack the mental models to recognize architectural catastrophe. You can prompt your way past syntax. You cannot prompt your way past a fundamental lack of engineering intuition. The hype told us that learning to code is dead because language is all you need. The data just proved the opposite. To truly master the vibe, you still need to understand how the machine thinks.

@ChaseLochmiller @OpenAI GPT-6 Astra, trained on ~100K+ NVIDIA Grace Blackwell NVLink72. From ChatGPT to o1 to Astra in 4 years. AGI has arrived. Congratulations @OpenAI team. 400K GPUs coming online next.

I'm increasingly confused why people think "agent memory for enterprise/teams" is a real product category that makes any sense at all

@BrianMRey Me too! Do you have any suggestions for me skills wise? I am not having it use any skills right now, don't want to mess it up.

万字长文 | Pi Agent 零基础实操：从打开终端到完成第一个真实任务

What techniques do you use for making AI-authored PR's easier to review? - I.e. use /show-me by @dexhorthy for easy-to-read diffs - Describe the changes in pseudocode - Mermaid diagrams - Evidence of TDD The wildest version of this I've heard is @dctanner's agent, who uses Playwright to record a video of the frontend feature working with text-to-speech over the top. So the agent literally talks you through its change.

🔥推荐：image2 + GPT-6 Astra 做 PPT是目前测过最快也最好看的一条路！ 核心不是“让 AI 直接出 PPT”，而是先出整页视觉稿，再精准拆成可编辑文件。结合文稿+配图时，传统 AI PPT 最容易丑、乱裁图；这条工作流刚好打在痛点上。 关键操作： 1️⃣先用 ChatGPT 把文稿聊成清晰的 PPT 文字结构和每页重点 2️⃣把文字 + 需要插入的配图交给 Image2，按页生成整页 PPT 图，不满意就重出，直到视觉统一、高级 3️⃣用 GPT-6 轻度版把每张图转成 PPT：提示词写清“按图层拆分、配图尽量原样保留，文字/流程图/框架全部可编辑” 4️⃣得到一套既好看又逻辑清楚、组件可复用的 PPT。GPT-6 拆层很准，基本不用手改 注意：转 PPT 比较吃 token，但效果比 Luna、Terra 稳很多。适合要“既要审美、又要能改、还要快”的场景。 #AI做PPT #GPT6 #Image2 #Codex #效率工具 #职场干货 #PPT技巧

I've updated and cleaned up my Starbase LA permit map to better visualize the regulatory filing activity since the Spacex annoucement. 3 total permits filed so far. The 20 square points on the map represent proposed soil borings to depths of 100'-230'. This permit is approved and work is ongoing. (Originally there were 25 borings permitted, but the permit was later amended to just 20. There was also a second permit filed for additional borings, but it was cancelled the next day). The 19 small white dots on the map represent proposed test pits included in the 3rd permit application. This one was filed last week and is pending approval. It's different from the first 2 permits as it involves bucket excavation and sampling to a depth of just 10 feet. No actual construction permits have been applied for yet. This is all for planning and design. In the map, you can clearly see a potential construction pattern and I've color coded those accordingly. Although this map is currently updated, I've seen these plot locations change before and they may change again. It does at least give you another sense of scale for this infrastructure. It also shows how they are having to work around the private land. Final note - I've seen plenty of X posts talking about the dredge barges and support equipment currently in Freshwater Bayou. I've mentioned this before, but it has nothing to do with this project. The bayou south of the locks is constantly silting in and the U.S. Army Corps of Engineers conducts periodic dredging to maintain the 16 ft required depths. They do this every few years and now they are doing it again. Dredged material is pumped into the gulf on the west side of the channel (always west side due to the counter clockwise gulf loop current so it doesn't fall back in). It's not going on Starbase. I'll probably do another post explaining this since it is getting so much attention.

🚀 Meet MiniCPM5-2B, a 2B-parameter language model bringing high intelligence density to the edge, now open source! It ranks #1 among open-source models under 4B parameters on the @ArtificialAnlys Intelligence Index, with a score of 23. It also scores 20 on the Agentic Index, bringing an early form of general-purpose agent capability to the edge. Across 34 benchmarks, MiniCPM5-2B achieves an average score of 53.9, covering coding, math, long-context understanding, tool use, and agentic tasks. And this release goes beyond the model itself. We’re opening up the data, training recipes, and RL stack behind MiniCPM5-2B. 🤗 Hugging Face: https://t.co/FZOMTZhBjq 💻 GitHub: https://t.co/2L0I8bYv8f Modelscope: https://t.co/WRlPNEAzgu Web: https://t.co/gRSu65FeZL

My dear friends, I am happy to report the publication of the most important paper in my life to date (we have several great papers coming out but this is very special). Tomorrow, I will present this paper for the first time at the Nature AI Healthcare in Paris and will post a longer post on this story and its broader implications for how to conduct clinical trials. Please read it and comment on it. Many thanks to the great co-authors of the study and everyone who contributed. Many thanks to the many reviewers (friendly and unfriendly) for spending so much time and helping make it better. Link in the comments.

It's never been a better time to own product/engineering at a non technical company. Like On shoes or Lululemon or American Eagle or something. I literally suspect that AI has changed the topology of non-tech companies more than it's changed the topology of highly technical businesses. At a lot of retail and media companies there's a very dominant Chief Digital Officer, head of e-commerce, CRO, whatever who owns the ultimate PnL for the business. They'll own a wide swath of marketing, e-commerce, site experience, customer support, CRM etc. but it hasn't historically been consistent that they own engineering (often reports to IT). There was historically a constant internal war to get these engineers to do more, competently. Unlike at a tech company, if you're selling women's shoes nobody cares about the software engineers. As a result these eng teams were significantly underfunded and often had much lower quality talent, specifically even if the floor was fine the ceiling was heavily limited - very few MIT and Waterloo grads on the tech team. Companies that bucked this trend (Netflix comes to mind) often dominate their markets with technology. But any engineer is hard to hire, even one who went to Tijuana Tech, so they've had to hire a lot of remote talent. That gets acceptable talent in the door but trades off intensity. This all was enough of a specialized headache that again eng often doesn't report to the PnL owner. But the shame of it is that they really should be the same unit. A better site, better data for targeting, a super slick retail support experience or whatever all seriously move the needle. So the functions were separate for convenience but it wasn't optimal. I think that the ability to pay a nominal fee for very reliably Very-good-not-necessarily-great engineering is going to make a very big difference for these companies and it'll be a big win for the ones who move engineering into their PnL chains ASAP. Will probably lead to higher purchases and Average Order Value and I suspect will reduce a lot of frictional costs from people constantly needing to yell at middling motivation engineers who don't report to them. Prob a win for consumers as well. It's also gonna help with simply coming up with ideas on what to do. You don't become CDO of Vuori because of your insane tech innovation skills but a lot of the best conventional ideas will also get suggested and prioritized by AI.

we have conjured an alien from a stargate

Moving your Coding Agents to the Cloud (ft. Codex Desktop & GPT-6 Astra) https://t.co/vzXiW8XGiE 00:00 The limitations of running agents locally 01:13 Parallel development and Git worktrees 02:03 Moving your coding agent to a remote machine 02:25 Introducing Railway Cloud Agents 02:52 Installing the CLI and signing in 03:03 Configuring your project and default agent 03:42 Bringing your agent skills 04:24 Connecting the Codex desktop app 04:48 How SSH configuration works 05:10 Adding the SSH connection in Codex 05:39 Creating a remote project and starting a chat 06:09 Long-running tasks and remote execution 06:26 ChatGPT mobile access 07:03 Using Railway from your coding agent 07:15 Orchestrating multiple cloud agents

next mise release is going to have my dream dotfile management setup I've wanted for years. you just tell it to track a file and it'll automatically sync up/down to a git repo. Before, I had kludged together a personal setup with launchagent/rcm but this mise implementation is way easier to use. mise bootstrap dotfiles origin set jdx/dotfiles mise bootstrap dotfiles track ~/.zshrc basically it's like a per-file dropbox backed by git. there's an option to have `claude -p` or whatever come up with commit titles.

How well do agents use test/verification techniques? https://t.co/7E31xiS2H0 https://t.co/schArJqkax

How to build a one-person $1M company with Grok Bot

I used GPT-6 Astra to get AAA PC games running on my iPad mini! Skyrim, Batman: Arkham City, Hades, and Age of Empires II. The actual Windows versions, running locally with touch controls. No streaming. Demo below and I'll share a repo and setup guides soon! https://t.co/zz8ze2R6Qn

agent-browser can now record at 60fps Just add --fps 60 https://t.co/xgjXjqUE1t

Never gonna give you up Never gonna let you down Never gonna run around and desert you Never gonna make you cry Never gonna say goodbye Never gonna tell a lie and hurt you Thanks for reading. We will do a global reset of the usage for all paid subscriptions so that you can keep enjoying Astra after burning through all of it doing fun 3D modeling in blender. The work week is about to start. Lands around 6pm PST today.

LATEST: Our revenue estimate for the AI economy reached $229 billion annualized by the end of August – up 3.5x in one year. https://t.co/lMV918P6id

0.9.0 is here, and it brings the most wanted herdr feature: all your machines running herdr, in a single client 🎉 control your agents and projects across local and remote machines without jumping between terminal tabs. https://t.co/afmHCbu7Pr

Recently found this repo by @adithya_s_k I am telling you this freaking goldmine !! https://t.co/uEWe0yMues

In the past 3 months I've tried: @conductor_build , @orca_build , @aarondfrancis Solo, @t3dotcodes , @MulticaAI @marcelpociot's @getpolyscope is the only tool I find myself getting back to. COW is so much faster than git worktree. The optional Reviewer agent is spot on.