jc

I did a thing. Native LTX model text to video app for mac. Open source and release available for download now. github.com/james-see/ltx-… Throw me a star or 3 if you like it. I was tired of ComfyUI failing on fp8 errors.

@LLMJunky @cursor_ai composer 1.5 was already really good for all my light updates use cases, going to try 2 tonight. Its funny they didnt do 2.1 or 2.3. or and bs with naming it pro or composer mini or lite bs. How dare they have sane naming conventions.

Nice. @cursor_ai just dropped their new "Glass" alpha, and they're leaning heavily into the simplified coding GUI trend that's been blowing up lately. First impressions are really positive. And just look at how insanely fast Composer 2 is. First impressions? Drop yours 👇

Im starting to #stablemax all my developed apps. Im pinning the versions of all dependancies, including the OS version, the zip code, and the ip address of the user. If you move from your basement to use my apps at a coffeeshop, tough luck. Don't like it, consider asking me to make a pro version that actually would then justify support.

@basedjensen And cursor is Zoroastrian

Openai is sunni and anthropic is Shia

@anemll Thank god i opted for 4TB nvme then for my m4 max 128gb ram.

M5 Max SSD test. SSD is Memory.

The winning move is making your own models when you are an IDE agnostic to any one frontier ai lab.

Just cut off a student driver. welcome to the real world nigga

@AlchemyAmerican “Vindicating his work” is english a second language of the poster? Makes no sense.

One of the best documentaries I’ve ever seen…

@GergelyOrosz Wrong

I am hearing tons of complaints from Cursor customers at enterprise companies: A silent change put almost all models Cursor uses behind Max mode. Devs who used to manage to “spread out” monthly credits over a month see all of it used up in 1-2 days. Are furious + switching.

"The Onion’s Exclusive Interview With Sam Altman" theonion.com/the-onions-exc…

@maxasteele @joeflood Or best. DC full of this archetype unfortunately. The other archetypes already covered by other bars.

they're calling it the Worst First Date Option In DC History

@akram_learns @amritwt Explain with examples. Ive refactored shit unmaintained github repos into updated organized and working. And that was last october using @cursor_ai

@amritwt planning looks great until you try complex, messy systems

absolutely none of the terminal coding agents have even matched the planning mode from cursor they have done something magical there

@amritwt Correct.

@Sam_Mindset @Polymarket Look up online social networks science. Dislike buttons do work. Censorship is not having the button in the first place. You really think “you can like this or nothing” is not stifling? You probably believe in participation trophies too.

@Polymarket A dislike button won’t improve the platform—it’ll just weaponize dogpiles. People won’t use it to filter bad content, they’ll use it to bury opinions they don’t agree with. Congrats, we just reinvented censorship but made it crowd-sourced.

BREAKING: 𝕏 rolling out dislike button over the next 30 days.

@Polymarket Its about time.

White Boy Summer is coming

@livingdevops Ill rely on linux chmod and user jails and groups and built-in rbac in linux thank you very much.

🔥Breaking: NVIDIA just open-sourced the guardrails AI agents should have had from day one. It’s called OpenShell. Announced at GTC yesterday. Your coding agent currently has access to your terminal, files, AWS keys, and network. OpenShell fixes that. What it does: - Filesystem locked at sandbox creation - Network blocked by default. - You whitelist what’s allowed - API keys never touch the filesystem. Injected at runtime only - Policies defined in simple YAML One command to sandbox Claude Code, Codex, or Cursor. The architecture runs a full K3s cluster inside a single Docker container. No separate Kubernetes install. Adobe, Atlassian, Cisco, CrowdStrike, Salesforce are already integrating it. Most teams solve agent security at the application layer. OpenShell solves it at the infrastructure layer. GitHub repo link in comments.

I hate those that speak for the world like its true. Bad faith and bad framing. Many people knew we were helping them train OCR etc. but what could we do. We got their services for FREE. Free means you are the product. We learned this during OG Facebook era. Maybe you were still in diapers or something? Or just larping as an influencer post now. Either way. Blocking this content to negate this interaction. Peace

Right. I use manus to generate FULL slide decks that would have taken me hours of research. End to end, 10 minutes later, I have a non-AI looking professional slide deck I use directly at work. $60 a month best bang for buck. Chatgpt and others do not compare at all in terms of the quality of output specifically. They nailed the delivery. They were one of the first to spin up their own compute / machines for the task and show the user the machine use. It works for me better than the others I also have accounts with.

@jc50000000 @DataChaz @karpathy Manus pricey for what they are doing lol. Not that deep

THIS is the wildest open-source project I’ve seen this month. We were all hyped about @karpathy's autoresearch project automating the experiment loop a few weeks ago. (ICYMI → github.com/karpathy/autor…) But a bunch of folks just took it ten steps further and automated the entire scientific method end-to-end. It's called AutoResearchClaw, and it's fully open-source. You pass it a single CLI command with a raw idea, and it completely takes over 🤯 The 23-stage loop they designed is insane: ✦ First, it handles the literature review. - It searches arXiv and Semantic Scholar for real papers - Cross-references them against DataCite and CrossRef. - No fake papers make it through. ✦ Second, it runs the sandbox. - It generates the code from scratch. - If the code breaks, it self-heals. - You don't have to step in. ✦ Finally, it writes the paper. - It structures 5,000+ words into Introduction, Related Work, Method, and Experiments. - Formats the math, generates the comparison charts, - Then wraps the whole thing in official ICML or ICLR LaTeX templates. You can set it to pause for human approval, or you can just pass the --auto-approve flag and walk away. What it spits out at the end: → Full academic paper draft → Conference-grade .tex files → Verified, hallucination-free citations → All experiment scripts and sandbox results This is what autonomous AI agents actually look like in 2026. Free and open-source. Link to repo in 🧵 ↓

@lambatameya @edwinarbus @cursor_ai “Finally”? It been that way since last summer.

@edwinarbus @cursor_ai the boost is context. cursor can see your entire codebase, your project structure, your dependencies. it's not that the models got smarter. it's that they can finally see what they're shipping into.

Matt Maher tested frontier models in Cursor v. other harnesses. Cursor boosted model performance by 11% on average: Gemini: 52% → 57% GPT-5.4: 82% → 88% Opus: 77% → 93% His benchmark measures how well models implement a 100-feature PRD. @cursor_ai consistently outperformed.

@sck_trade @edwinarbus @cursor_ai Sluggish? Lol. any data to back that up.

@edwinarbus @cursor_ai That, and VS Code feels like a dream compared to how sluggish Cursor gets. Good luck running multiple projects at the same time in multiple Cursor instances. I have a $15k ai desktop rig. That should not be a problem.