Dan Jasnowski 🕊

I got 2 exact same AI generated contributions and it is intimidating how much code they are generating. The one on the left is the AI generated PR code, the one on the right is the actual fix needed. The bug is SUPER easy while the agent is likely mislead by the context provided in the issue which is leading to 100x more code to fix it. My change was also AI generated, it's just the way you phrase your request. It's not a surprise Garry generates 100k LOC a week

Fantastic open.spotify.com/track/3LnxvF71…

🚨BREAKING: Someone just built a 12MB binary that gives AI agents full browser control. It's called PinchTab. No Playwright. No Puppeteer. No bloated dependencies. Just a plain HTTP request and your agent clicks, types, and navigates like a human. → 13x cheaper than screenshots (uses accessibility tree instead) → Bypasses bot detection out of the box → Runs multiple Chrome instances in parallel with isolated profiles → Works with Python, TypeScript, Go, or any language that speaks HTTP → Installs in one command Most browser automation tools weigh 500MB+ and need Python environments, npm nightmares, and Docker just to say hello. PinchTab is 12MB. One binary. Zero dependencies. This is what the browser control layer for AI agents actually looks like. 100% Open Source. MIT License.

@Qromerolauro Loving the website design aesthetic.

We just built Figma for Claude Code > Select any element on your local front-end > Edit it like you would in Figma > Apply the changes with Claude Code This is not a demo or waitlist. Try it today.

@dani_avila7 What would you suggest when I want to search over documents (scanned docs + legal docs, etc.)? Still agentic search? Or RAG?

We saw exactly this in my previous startup RAG + vector DB gives decent results, but agentic search over the repo (glob/grep/read, etc) consistently worked better on real-world codebases. We even pushed further: RAG + embeddings + AST + tree-sitter. The quality was excellent But exactly as @bcherny mentions: staleness and privacy, you need continuous re-indexing, and all the code and embeddings must live on your servers. In practice, fast models + bash-style agentic search ended up outperforming general RAG search, even if it requires more tool calls This is what we built 👇

We just shipped the full deep dive on how we rebuilt webhooks at @dodopayments and pushed delivery reliability to 99.99%+ with sub-500ms P50 latency. The uncomfortable truth: We were “only” losing ~0.3% webhooks. At 100K payments/day, that is ~300 customers/day who paid and got nothing. Root cause was not network failures. It was process failures. The classic flow is fundamentally fragile: DB commits ✅ App fires webhook from memory ❌ Deploys, SIGTERM, OOMs, restarts = silent drops. The mental shift that fixed it: Webhooks are derived data. So the event must be created durably first, then processed by systems built for durability. What we built (5 layers): 1. Postgres trigger writes a minimal domain_events row in the same transaction as the payment update 2. Sequin reads WAL via logical replication (no polling) and publishes in commit order 3. Kafka buffers durably + gives replay (we keep 7 days) 4. @restatedev runs the workflow with durable execution + durable retries (crash mid-flight is fine) 5. @SvixHQ handles last-mile delivery with stable idempotency keys Two details that mattered a lot: → One idempotency key per event, not per attempt (generated once, journaled, reused on replay) → Payload enrichment at delivery time (merchants want current state, not stale snapshots) Results: Before: ~300 lost/day at 100K webhooks/day After: <10 lost/day, and those are always merchant endpoint issues Link in the first reply.

This has been said a thousand times before, but allow me to add my own voice: the era of humans writing code is over. Disturbing for those of us who identify as SWEs, but no less true. That's not to say SWEs don't have work to do, but writing syntax directly is not it.

@taylorotwell @YokoHyper @heytjmiller Have you looked into Pinecone?

@YokoHyper @heytjmiller As part of AI SDK I added Postgres vector stuff into the framework, a Similarity Search tool, as well as file vector store support for OpenAI and Gemini

Laravel AI SDK demo at Laracon India then I’m tagging a public beta. 🫡

I'm Boris and I created Claude Code. Lots of people have asked how I use Claude Code, so I wanted to show off my setup a bit. My setup might be surprisingly vanilla! Claude Code works great out of the box, so I personally don't customize it much. There is no one correct way to use Claude Code: we intentionally build it in a way that you can use it, customize it, and hack it however you like. Each person on the Claude Code team uses it very differently. So, here goes.

@PaulLeonardMorg Next year! I promise!! As always, keep up the great work. See ya next year. 🥹

@danjasnowski It makes me sad that there’s not so much happiness or happy pills in your life this year ☺️

Another year of #SpotifyWrapped... another year of @PaulLeonardMorg being at the numb-- wait 3RD PLACE?!?! OH NO. There's a glitch in the matrix. Forgive me, Paul.

@PaulLeonardMorg x.com/i/status/19962…

@danjasnowski Which one?!

I scored 1930 on Jingle challenge #537! I finished in 01:55 and placed in the top 35.9%, can you beat me? jingle.rs 375 🟡 4 🟡 544 🟡 1000 🟢 7 🟡

@nikitabier I found the combination of these two tweets hilarious 😂

Thanks for all the suggestions everyone

If you get upgraded on a flight but your girlfriend/wife doesn’t, do you downgrade and both suffer or do you take the upgrade? I need an answer in 5 minutes.

@JackEllis @adamwathan What is this referring to

@adamwathan Biggest grift in history.

Are there any other open source apps we could fork and move the tabs to the side then sell for $610m? 🤔

@marshal Looove the color and the car itself with the Lamborghini Huracán EVO RWD. ☺️

I love exotic cars. I hate losing money. Most cars lose value fast. But with exotics and collector-grade cars, if you buy the right spec at the right time, you can often sell for what you paid—or even turn a profit. Here’s how I’ve done it (and why I’m not bleeding cash doing what I love): My best example: 2018 Porsche 911 GT3 Bought right after I sold my company. Manual, bucket seats, carbon brakes, rare PTS British Racing Green color. I just sold it for a $50k profit after 12 months of ownership. That’s a 22.2% return—while enjoying it for 1,700 miles. More examples of my cars: ⸻ 2022 Lamborghini Huracán Evo RWD I’ve had it for 3 years. I spec’d it new from factory. Still worth about what I paid. No loss. No regrets. ⸻ 1998 Porsche 911 Carrera S Last year of the air cooled Porsche’s. Blue interior, wide body, hardback seats, 6 speed. Will continue to appreciate. ⸻ 2016 Porsche Cayman GT4 Bought for $98K. Sold for $102k after 10 months and 1,500 miles ⸻ 2010 Lotus Exige S260 Bought for $86k Sold for $81.2k after 24 months. Multiple track days. Average loss of $200 a month. ⸻ Yes—there’s still gas, insurance, and maintenance. This isn’t a “get rich off cars” strategy. It’s just my way of enjoying incredible machines without losing 6-figures to depreciation. I don’t treat cars as investments. I just hate burning money. This doesn’t work with all exotics. Some just keep depreciating. I follow enthusiast sentiment around various cars. If you spec smart and buy near the bottom of the depreciation curve, you can drive a dream car and keep your money intact.

@tacobell can I get free tacos if I fix this bug for you? I cannot order my tacos. :(

@AdobeCare There is literally nothing you can do to help unless you rebuild the software with those changes I addressed. Oh well.

@danjasnowski Hi there, sorry for the trouble. Please send us a DM with more details about the issue with your Adobe ID so that we can check and help. ^Rohan twitter.com/messages/compo…

Why does @Adobe Acrobat suck? Let me count the ways. After installing... 1. It forces you to restart your machine. 2. It attempts to install a Chrome plugin 3. "Open PDFs with Adobe Acrobat?" x3 4. "No Credit Card Required." Unless you want to try the free trial.