Sabitlenmiş Tweet
Akarsh G
2.4K posts
Akarsh G
@akarsh_ghale
Building @CygnusAIPrime | Deploy your first OpenClaw AI Employee | 🌐https://t.co/fCvuJFry5F
Katılım Şubat 2018
273 Takip Edilen157 Takipçiler
Story time
so some company ran a marketing campaign and gave money to all the big creators on X to comment under their post. Someone screenshot it and posted it, and now it looks like paid marketing, which isn’t allowed on X unless you disclose it. Nikita saw that post and removed them from payouts. Tomorrow is going to be crazy.
Srajan@_Creation22
Software engineering are going to get replace in next 6 to 12 months ahh replies 🤣🤣🤣🤣🤣
English
@ctatedev Hey Chris,
Can you take a look at this PR?
github.com/vercel-labs/ag…
English
Introducing json-render devtools
Inspect Generative UI in real time
→ Spec, State, Actions, Stream tabs
→ DOM picker maps clicks to spec keys
→ Catalog browser for your components
→ Shadow-DOM, tree-shakes in prod
Drop in <JsonRenderDevtools />
React, Vue, Svelte, Solid
English
@ctatedev Hi Chris,
Would you mind taking a look at this PR that’s been hanging for more than 2 weeks now?
Thanks
github.com/vercel-labs/ag…
English
one of the nicest effects of portless
i *always* know which app i'll get when i hit enter in the address bar
and *never* think about ports
no "what port is this on"
no conflicts
worktrees don't fight each other
everything has a predictable URL
other devices on my network can hit it
really glad it's useful to people, and grateful for all the incredible contributions the community has made to make it better
Michael@michael_chomsky
I genuinely unironically think @ctatedev from Vercel Labs saved local development with Portless. Genius.
English
@ctatedev Hi Chris,
Can you take a look at this PR? It’s can hanging for more than a week now.
Thanks
github.com/vercel-labs/ag…
English
Add it to your React or Vanilla web app
github.com/vercel-labs/wt…
English
wterm trades peak performance for usability, accessibility and native browser features
And also - perf is *good enough* for many use cases
Chris Tate@ctatedev
Introducing wterm (“dub-term”) A terminal emulator for the web → DOM rendering — not canvas → Select text, copy/paste, ⌘+F, a11y → Dirty-row tracking, 24-bit color, themes → WebSocket transport with reconnection → Zig core compiled to ~12 KB WASM → just-bash, local, SSH
English
Use @composio and create MCP with your integrations. Tokens or API keys never touch your instance. All at the network layer.
If you are interested you can get a secure cloud hosted at @CygnusAIPrime . I am building that app to make it easier to manage openclaw and let us handle the security and updates.
I can give you one month free trial just let me know. You can even connect your OAI sub like plus or pro.
English
It’s April now
OpenClaw with docker sandbox, logging mitmproxy firewall and Clawvisor and you are good to go
The days of “it’s insecure” for OpenClaw are over
Peter Steinberger 🦞@steipete
That was the case in December. 4 months and thousands of work hours later, we have a great security concept; you can go all yolo, use a sandbox (Docker or OpenShell), there are allow-lists and per-access exec allow/deny prompts. There’s hundreds of security researchers that pen-tested it.
English
@ctatedev Can you please take a look at this PR that’s been waiting for more than a week now?
Thanks
github.com/vercel-labs/ag…
English
Batch is so good
agent-browser batch "open https://app.localhost" "wait 3000" "snapshot"
English
English
New command: agent-browser skills
Cached skills go stale when the CLI updates. Now the CLI serves skill content at runtime.
One thin skill installed via:
npx skills add vercel-labs/agent-browser
Evals show agents pick agent-browser more often and use it correctly.
English
I don't understand this. 5.4 handles complex, longer-horizon task way more reliably than Opus 4.6 as per my experience. Don't understand how it sucks here.
Although Opus has greater creative output but it is fraught with bugs and mistakes which makes it unreliable for production.
English
2 days ago GPT 5.4 ran for 12 hours proving 107 different theorems while I was sleeping.
Yesterday it ran for 14 hours formalizing Shor's Algorithm. I think that needs another day or two to finish.
METR@METR_Evals
We ran GPT-5.4 (xhigh) on our tasks. Its time-horizon depends greatly on our treatment of reward hacks: the point estimate would be 5.7hrs (95% CI of 3hrs to 13.5hrs) under our standard methodology, but 13hrs (95% CI of 5hrs to 74hrs) if we allow reward hacks.
English
Excited to launch our brand new photo editor. Now available on iOS, with Android coming soon. Lots of good quality of life updates and more improvements to come.
English
@pashmerepat On cygnus-prime.ai you can get your fully secure, cloud hosted openclaw🦞 instance and connect your codex sub without having to go through any CLI stuff.
English
Exactly one year ago, this is what models could score in BrowseComp.
Now Claude Mythos hit 86%.
Lisan al Gaib@scaling01
Claude Mythos is insanely token-efficient
English
English
@agentmail @browser_use I think we should move away from browser emulation, agents don’t need UI/UX. everything should be accessible from CLI
English
AgentMail is now a native integration in @browser_use
Every browser session gets its own inbox. Sign up for services, verify emails, extract OTPs, and maintain a persistent identity across sessions.
No human in the loop.
English
@garrytan @conductor_build Hey Garry, we made hosting openclaw super easy and we manage the updates, security with AES-256 bit encryption. In case you’re interested please do give it a try. You never need to mess around with cli or config.
Would really love to get your feedback. cygnus-prime.ai
English
I’m on a United flight with their legendarily bad WiFi that makes internet unusable on my MacBook Pro especially for heavy multi-worktree @conductor_build sessions
But now I have a new love: building in multiple topics in Telegram group with my OpenClaw bot
It feels amazing
English
Only issue I have faced is gpt-5.4 doesn't realize stuff until you nudge it. like if I have created a MCP server, i need to explicitly tell it to lookup the MCP tools available.
Another issue is it doesn't automatically use skills until I say tell it to.
These are some issues I hope are fixed with the new update, haven't had the time to test.
What's been your experience so far with this?
English
How to fix your OpenClaw:
Option 1: panic, search Twitter and Google for hours, read the 48th blog post, and keep spinning
Option 2: switch to openai-codex/gpt-5.4 and just try it
English
@Elaina43114880 Holy sh*t! Not one text mistake in sight. This is f*cking crazy!
English
@openclaw We just rolled out this update to our running instances.
Create your own secure OpenClaw instance without the hosting headaches at cygnus-prime.ai
English
OpenClaw 2026.4.5 🦞
🎬 Built-in video + music generation
🧠 /dreaming is now real
🔀 Structured task progress
⚡ Better prompt-cache reuse
🌍 Control UI + Docs now speak 12 more languages
Anthropic cut us off. GPT-5.4 got better. We moved on. github.com/openclaw/openc…
English
A lot of the decacorn AI agent cos and labs other than OpenAI are trying to kill OpenClaw or replace it
However: I think community and open source is too strong and the Apple II moment will actually happen for OpenClaw itself, not for some corporate closed source solution
English
"There’s a big wave coming" #theres-a-big-wave-coming" target="_blank" rel="nofollow noopener">mtlynch.io/claude-code-fo…
English