Extracting one archive is easy. Extracting 20 is a waste of time. Today we launched EasyUnZip for Windows 🚀 Select your ZIP or RAR files in File Explorer, press Alt + E, and extract them all at once. • No file uploads • No clutter • No subscription Try it for free: easyunzip.app

take on tibo's 28 days codex improvement challenge..... we have got more resets than useful improvements.

when claude code hits the limit and you jump to codex, how do you hand over the project? i paste a 5 line summary. feels caveman. what are you doing

claude has made a judgement call on sol vs muse Codex Sol (6.1, medium effort) Reliability: it finished every task I gave it today. It only stopped once, when it hit its usage limit. Instructions: it follows a brief closely and reports in the requested format. Muse (Wonder Full, high effort) Code quality: on the one task it finished today the work was careful. It wrote the failing tests first, covered edge cases like old saved notes, and explained each item clearly. Reliability: this is its weak point. It kept stalling on 5-minute “no first model event” timeouts

hows the usage limit now in codex? planning to get pro subscription. moved to claude because of unusable usage limits on codex.

every week there's a new "best coding model ever". my weekly limit, watching me try all of them:

Very important question right now… Whats eating grok bot usage?

vibe coding is fine. shipping code you never read to people who paid you isn't. if your app eats someone's files, "claude wrote it" is not an answer

next build after easyunzip: qkr.si a remote mcp server, so claude, chatgpt and copilot can all call the same tool. domain is live on cloudflare. the server isn't yet. building it with codex and claude code, will post the ugly parts

tibo just reset everyone again. me, who already rationed my week like it's 1943:

tokens processed is a weird thing to brag about. 36.5 trillion here, 25 trillion there. my agent re-reading the same repo 40 times also processes a lot of tokens. tell me how many tasks got finished

how i used claude code in august: one session open all day, type "continue" after every break. how i use it now: fresh chat per task, 5 line summary before i step away, codex for the boring stuff. same plan. way fewer "usage limit reached" screens

fable 5.5 "drops today", day 3. me still refreshing the claude model picker:

question for people paying for more than one: which $20 plan actually gets you through a day of real coding? pick one:

lesson from 50 replies on tuesday: my 7 word reply to @openaidevs did 1,784 views. my 3 paragraph reply in the same thread, 4 minutes earlier, did 188. on x the reply that sounds like a person beats the one that sounds like a blog post

lesson from 50 replies yesterday: my 6 word reply to @openaidevs did 1,727 views. my 3 paragraph reply in the same thread, 4 minutes earlier, did 183. on x, the reply that sounds like a person beats the one that sounds like a blog post

quick tip for claude code, codex, cursor, whatever: commit before every agent run. check git diff after. if a one line fix touched 14 files, don't argue with it. git reset and ask again smaller

myth: opus 5.5 got nerfed. reality: the before/after tests going around show the same output. it just takes a few more prompts to get there. so it's not dumber, it's more expensive. kinda worse honestly

built easyunzip.app because right click > extract all > pick a folder got old fast. now i select the zips in explorer, press alt+e, done. it's on the microsoft store now. $9.99 once, no subscription

quick tip if your claude code limit is draining faster this week: don't type "continue" in a chat that's been idle for an hour. the cache has expired, so you pay to re-read the whole thing. before you step away, ask for a 5 line summary. start a fresh chat with it when you're back

token counts are the new vanity metric. t3 code: 36.5 trillion. gemini 4 argon: 25 trillion in 72 hours. coding agents re-read the same repo every turn. the number i'd brag about is how few tokens a finished task took