Stas Persiianenko

755 posts

Stas Persiianenko

@itspers

Kharkiv เข้าร่วม Ağustos 2010

946 กำลังติดตาม64 ผู้ติดตาม

@neilhtennek I read it and don't understand how to use it. I should pair each session manually when I need to or what? Light right now I can tell openclaw to go to tmux and tell Claude/Codex/Cursor to do something, how this alternative supposed to be used?

English

kenneth@neilhtennek·8h

plz go use this thing i built <3 code.claude.com/docs/en/channe…

English

212

10.6K

Stas Persiianenko@itspers·13h

@leerob Can it run in cloud with computer use?

English

Lee Robinson@leerob·15h

Composer 2 is here! Fast, intelligent, and cost effective. It's our first continued pretraining run + even further scaled RL. Try it out 😄

Cursor@cursor_ai

We were able to significantly improve the model quality and cost to serve. These quality improvements come from our first continued pretraining run, providing a far stronger base to scale our reinforcement learning.

English

883

60.1K

Stas Persiianenko@itspers·2d

@neilhtennek Whatever use case I tried to invent for this Cowork - it says it runs in VM and can organise files in my folders..

English

kenneth@neilhtennek·2d

go use thing i worked on plz ty

Felix Rieseberg@felixrieseberg

We're shipping a new feature in Claude Cowork as a research preview that I'm excited about: Dispatch! One persistent conversation with Claude that runs on your computer. Message it from your phone. Come back to finished work. To try it out, download Claude Desktop, then pair your phone.

English

5.9K

Stas Persiianenko@itspers·2d

@rudzinskimaciej @morphllm @albfresco @mots_pod I already have logic of auto compacting in code, which calls now some cheap model via open router. If this model would be provided in open router - I would just try. But not by creating another account and writing new code...

English

Rudzinski Maciej@rudzinskimaciej·2d

@morphllm @itspers @albfresco @mots_pod Do a openrouter endpoint where you have someones model e.g. k2.5 or minimax2.5 but with your auto compaction so few % more expensive or whatnot similarly to oai endpoint with auto compaction

English

Morph@morphllm·2d

Introducing FlashCompact - the first specialized model for context compaction 33k tokens/sec 200k → 50k in ~1.5s Fast, high quality compaction

English

137

2.1K

207.6K

Stas Persiianenko@itspers·2d

@morphllm @albfresco @mots_pod Too much work to try. Would be cool if it was available on open router.

English

Morph@morphllm·2d

@albfresco @mots_pod its available via endpoint! docs.morphllm.com/sdk/components…

English

4.2K

Stas Persiianenko@itspers·2d

@felixrieseberg Please stop vibecoding all this and fix what you have. Fix claude code - you can add this endless wrappers for managers later.

English

Felix Rieseberg@felixrieseberg·2d

English

942

1.5K

17.3K

Stas Persiianenko@itspers·2d

@cramforce Can you explain how it relates to ai SDK?

English

Malte Ubl@cramforce·2d

Reasons why it makes sense to use chat-sdk.dev even if your agent only targets Slack: Lots of little details "just work": - Channel names are converted to clear text so your agent knows what people are talking about - Same with user names - Same in reverse: When the agent at-mentions somebody in clear text, that actually works - Link preview content, referenced posts, and images are automatically included in agent prompts, so that your agent has the full context - Agents love markdown, but Slack doesn't natively support it. Conversion from standard markdown to Slack-variant is automatic. - And that even works when using Slack's native append-only streaming API

English

138

12.5K

Stas Persiianenko@itspers·4d

@mathemagic1an @dominikkoch @mlejva So if i will ask to recommend me something to replace codegen, but not clickup, and somebody will ask why, and i will honestly anwser - i will be asshole?

English

Jay Hack@mathemagic1an·4d

As an independent third party here I can confirm you are being a bit of an asshole This sounds like something you should either contact them about directly or, alternatively, if you are unsatisfied, address directly on social media by talking about your experience with their HR. Vague posting and calling out their product is passive aggressive and not a good look for you. Otherwise: Daytona and Modal are solid offerings. Bleeding edge is just-bash depending on what your requirements are.

English

1.9K

Dominik Koch@dominikkoch·4d

hey gang what sandbox provider would yall recommend (not e2b)

English

27.6K

Stas Persiianenko@itspers·8 Mar

@manthanguptaa Trick is that vscode is not really IDE

English

Manthan Gupta@manthanguptaa·8 Mar

I love Cursor, but the latest update is making the IDE keep crashing, man! Can you all stop vibe coding the IDE?

English

105

12.6K

Stas Persiianenko@itspers·6 Mar

@ericzakariasson @gilesvangruisen You told this every week. And then just add more new semi-useful features and bugs.

English

eric zakariasson@ericzakariasson·6 Mar

@gilesvangruisen just wait

English

789

Giles Van Gruisen@gilesvangruisen·6 Mar

Devs: ugh cursor just keeps getting messier and messier Cursor:

eric zakariasson@ericzakariasson

should we ship this in cursor

English

1.6K

Stas Persiianenko@itspers·6 Mar

@msfeldstein But there are no filters, you basically spin up VM and agent on every PR?

English

Michael Feldstein@msfeldstein·6 Mar

"/create-automation send a notification to #-keybindings-guardian whenver a PR merges that modifies keybindings in vscode"

English

280

Michael Feldstein@msfeldstein·6 Mar

crazy thing is how easy it is to create these. I wanted to get notifications whenever someone adds or changes keybindings and it was more work to create the slack channel i wanted it funneled to than to create the automation

Cursor@cursor_ai

We're introducing Cursor Automations to build always-on agents.

English

3.3K

Stas Persiianenko@itspers·6 Mar

@mntruell That explains why they are so buggy. Future is here, human QA is already elite feature, which nobody can afford

English

Michael Truell@mntruell·5 Mar

Automations already run thousands of times per day inside our own codebase! They power self-healing CI, auto-approving PR flows, highly-compute-intensive security review, and a team-wide memory system. One small step toward a self-driving codebase...

Cursor@cursor_ai

We're introducing Cursor Automations to build always-on agents.

English

474

49.2K

Stas Persiianenko@itspers·5 Mar

@vincentmvdm @mainframe Can you explain second a bit more?

English

Vincent van der Meulen@vincentmvdm·5 Mar

our favorite cursor automations at @mainframe: 1. unslop: reviews against agents.md and calls out effects per "You Might Not Need an Effect" 2. generator-verifier: cursor critiques a pr against a plan on every push; cloud agent iterates until satisfied

Cursor@cursor_ai

We're introducing Cursor Automations to build always-on agents.

Manhattan, NY 🇺🇸 English

13.6K

Stas Persiianenko@itspers·4 Mar

@leerob Is it possible to invoke debug mode somehow via prompt?

English

939

Lee Robinson@leerob·4 Mar

You can now use Cursor with 30+ ACP clients, including OpenClaw 🦞 This means complete access to Composer 1.5, codebase indexing and semantic search, and more! Here's an example with avante.nvim

English

739

125.9K

Stas Persiianenko@itspers·2 Mar

@JosephD @ericzakariasson It looks like some marketing strategy, whole Cursor team keeps positing some usecases of burning tokens for last weeks, prenteding everybody outside has unlimited plans or so

English

Joseph@JosephD·2 Mar

@ericzakariasson Imagine if the architecture didn't allow those bugs to exist in the first place. That 10 hours of token spend disappears.

English

1.9K

eric zakariasson@ericzakariasson·2 Mar

here's a trick to get long running agents to produce better outputs: ask it to add a feature flag in the codebase, with the condition that it should always file when its disabled and always succeed when enabled this essentially becomes red/green testing where you need the model to have a failing test, and then a passing i saw this first hand from @davidgomes when he fixed a inline diff bug with an agent that ran for 10+ hours

English

635

64.3K

Stas Persiianenko@itspers·26 Şub

@kentcdodds Do you have some unlimited sponsored plan, or just spending some crazy money? Or maybe it is really usable on small codebases?

English

Kent C. Dodds ⚡@kentcdodds·26 Şub

I've been doing this for a while now and it's awesome to kick off an agent to work on something before I go to bed. I then wake up in the morning and see bugbot and the agent iterating on the solution

Cursor@cursor_ai

Cursor can now automatically fix issues it finds in PRs with Bugbot Autofix.

English

7.8K

Stas Persiianenko@itspers·26 Şub

@ericzakariasson So what it was?

English

eric zakariasson@ericzakariasson·24 Şub

phenomenal day to be a cursor user. i think this is the most exciting thing we've launched in a while, and a defining moment for 2026

Kent C. Dodds ⚡@kentcdodds

Today is a good day to release stuff.

English

678

136.4K

Stas Persiianenko@itspers·25 Şub

@corbin_braun It's just too expensive to use for real work full time. Like I tried, it works, but I will not use it anyway.. And they will not make it work cheaper, cause whole idea is that we run expensive models, with computer use which burns tokens as crazy to make it work

English

305

corbin@corbin_braun·25 Şub

honestly. by the way, people are not even reacting to this new Cursor drop tells me two things: 1) 90% of devs don’t even pay attention / don’t care. let me just get my next paycheck. 2) If you are at the edges of new tech, you have a significant competitive edge. This drop is just mind boggling. Your canary in the coal mine for people who talk in tech is if they identify this as a big deal.

English

415

62.8K

Stas Persiianenko@itspers·25 Şub

@fredrikalindh yeah, its definitelly look like its vibecoded

English

362

fredrika@fredrikalindh·24 Şub

i joined cursor a month ago to build cloud agents - today we’re launching agents that can test and demo their work last week i shipped 60 PRs with them this is how i use them:

English

547

43.8K

Stas Persiianenko@itspers·25 Şub

@leerob To make this work on something more real than hello world react websites need docker in docker and devcontainer support.

English

Lee Robinson@leerob·24 Şub

Cursor just got a major upgrade! Agents can onboard to your codebase, use a cloud computer to make changes, and send you a video demo of their finished work. The latency of using the remote desktop is smooooth.

English

226

151

3.1K

510.8K

ค้นพบ

@neilhtennek @leerob @rudzinskimaciej @morphllm @albfresco @mots_pod @felixrieseberg @cramforce