⚫ Move Fast, Stay Curious ➤ Think in systems, build with intent ➤ Bias for action — clarity comes from doing ➤ Here to build what feels right
India
Joined September 2024
- Tweets1.1K
- Following626
- Followers378
- Likes12.5K
Pinned Tweet
Joined the Hackathon out of curiosity, and it turned into something special.
Big thanks to @WeMakeDevs, @cerebras, @Docker , @Meta and the team behind this amazing event.
🏆 Cerebras track winner – FutureStack GenAI Hackathon 2025 🏆
Big congratulations to @hpaptan7 🎉 - winner of $5000 at FutureStack 2025!
💡 His project: VoiceGraph – speak your automations and watch them come alive, powered by Cerebras for lightning-fast voice-driven execution and AI-powered debugging.
Project demo: youtube.com/watch?v=kzdXmUTY…
Let's Fucking goo!
@herdrdev is back with a banger! 🤯🤯🤯
Got early access to Rex by @superlogical 🎉
Sessions you can reconnect to from any device, and Claude Code or Codex can open tabs, split panes and read terminals.
Catch: beta is macOS-only, I'm on Arch. Linux is coming👀
@mitchellh @pearkes @almonk @dizzyup
Also, I wanted to push some changes as well on it , can i do that? I really want to switch between multiple claude subs like work or personal in a single session.🧐
@winglee
I lovee MyGo,
Something coming up soon, built on top of it =)
and yes ofcouse , opensource.
I created MyGo just three weeks ago, and it's already the best desktop GUI framework for Go, unbelievable things are happening thanks to Opus 5.5
mygo.egoist.dev
With my X account I can earn up to $9 a month. How much can you earn? creators.unfair.so/worth/hpa…
8.4% of vendors supporting short lived credentials is a specific number. this week i want to list my own static tokens and see which ones can be swapped
It's time to retire long-lived static credentials.
We surveyed 1,000+ SaaS vendors and found only 8.4% support secure auth. This is our call to action and how we're doing our part.
vercel.com/blog/death-to-lon…
my always loaded notes file:
• 37KB, so every session paid for it before i typed a word
• i moved the history out and kept the rules
• now it's 7.7KB
screenshot is both files side by side
my flaky fake agent:
• every other request fails with a rate limit error
• the retry says "Second try worked."
• it finds REAL UI bugs for free
the flaky one is my favorite
my GNOME login went grey after a pacman -Syyu:
• mesa was built for LLVM 22
• llvm-libs was 23
• one mesa reinstall from cache fixed it
screenshot is the version check on both
Tracing for agents:
• every step on one timeline
• cost and tokens per step
• click a step to see its input and output
different kind of tracing i believe but so much fun!
my agent panel runs on an old ThinkPad i can't physically reach:
• systemd user unit, linger on
• a crash or a reboot brings it back without me
• screenshot is the status line
Claude wrote formal proofs for 100+ open source libraries, down at the RISC-V assembly level.
• it also found bugs that years of tests missed
• i want to read how they check the proofs first
• then try the idea on a tiny library of my own
good read from @AnthropicAI
