BSKR

Claude’s asshole

@bridgemindai Planning then using /goal, even better

GPT 5.5 xHigh in Codex with plan mode is the best coding workflow I've used. Make it plan first. Review the plan. Then say "implement the plan." The results are night and day versus just prompting raw. It creates a structured test plan, lists assumptions, and then executes exactly what it outlined. No guessing. No hallucinated files. Just clean implementation. If you're using Codex without plan mode, you're leaving half the intelligence on the table.

@sama @jakeottiger Cool, but more goblin merch would be cooler

@jakeottiger that does sound pretty cool

@elonmusk sir when Grok CLI, tired of Claude code and codex..

Damn is Claude Code becoming mainstream?

@kiragu_maina_ @Claude I really like codex, but only on 20 plan hits limits in 2 tasks. But if they release 100 plan its probably worth it

@buskerrrrrr @Claude I always have a docs folder with specs and plans.. just incase claude hits me with a limit.. been thinking about paying for codex.. huge hype around it.. the last time i used it was last year and it was not that good

@AnthropicAI has a major outage and out of Codex messages. What do they expect me to do, write my own code?

@kiragu_maina_ @Claude Imma give it a shot, also need to have a backup for when shit like this happens

@buskerrrrrr @Claude There's gemini.. if u had already created the plans with claude.. I think gemini can do a good job.. albeit slower and with many iterations.

@kiragu_maina_ @Claude My ass getting fired for zero productivity today

@buskerrrrrr @Claude 😂... how dare they... we are thinking on a different level.. imagine typing out function by function..

Claude Code is absolutely cooking again. It feels like they fixed their issues with opus being retarded.

@cursor_ai Really impressed by Cursor lately

Use Design Mode in Cursor 3 to annotate and target UI elements in the browser.

Genuinely asking: what do vibecoders do when they hit rate limit?

new keyboard wdy think

@code_kartik @AnthropicAI It is so frustrating…

Used just 3–4 prompts and somehow I’m already at 41% usage? That doesn’t add up. Either the tracking is off, or the usage model isn’t as transparent as it should be. Anyone else noticing this? cc: @AnthropicAI

@daremarket 14

the number was 427. nobody got it. let's run it back: $200 double or nothing. guess a number between 1 and 1,000

@gordonmattey My just got reset, and now hit within 45 minutes. Maybe something weird going on with opus 4.6 with 1M context

@buskerrrrrr Also had the same thing happen to me.

Paying $100/month for Claude Max and hitting the rate limit in one hour is insane. This has never happened to me before. I usually run multiple sessions, sub-agents, and research workflows without getting capped. What changed @AnthropicAI? @DarioAmodei @jackclarkSF

Yeah it's bad, yesterday I had heavy coding sessions with 40+ commits and it barely touched the limit. Today I let it research a small new topic with 2 sub agents and the usage already jumped to 50%. Then just a few short messages completely burned through the entire 5 hour limit.

@buskerrrrrr @AnthropicAI @DarioAmodei @jackclarkSF 1h ? I said Hi and it jumped 15%, must be a bug on their side. yesterday I had 51 commits throughout the day in a project.. today a few "hellos" and it kills the 5h limit.

Love Claude and want to keep scaling with it, happy to share more detailed workflow data privately if it helps the team tune limits or docs for Max users. Any guidance would be huge. Thanks!