300k stars for a folder of AI instructions
mattpocock/skills is what Matt Pocock uses every day. small, plain, easy to tweak
I ran it on my 5 hardest tickets. it finished all of them then asked for more work
you using any skills yet?
there is an anti-OpenAI cult, and it’s because these X influencers have invested their money into the IPO and they’ll do anything to give them the best chance at cashing out. half this shit is just a bunch of adults whining at a computer
fable 5.2 costs 2x opus per token and can still be cheaper
on Anthropic's coding eval, Fable on High effort matches Opus 5.5 on Max for about half the cost per task. they still say start most work on Opus
price the finished task, not the token
https://t.co/ZPzI8f4FeE
@almokhtar_al OSEP labs and a run in Cologne is a proper day G. I'm in the AI coding tools lane right now, mostly Cursor and Claude tests. since you're coming from infosec, what's the first AI threat you're actually drilling, prompt injection or agents with too much tool access?
Introducing Claude Fable 5.2 and Claude Mythos 5.2, designed to tackle the most daunting technical challenges.
They’re capable collaborators on ambitious projects in many fields, and show improvements in areas like coding, reasoning, and vision.
We're giving away a Hogwarts Legacy custom wrapped NVIDIA GeForce RTX 5090 to celebrate its official release with DLSS 5 🪄
To enter:
🟢 Share this post
🟢 Comment #RTXON
Terms and Conditions: https://t.co/45T3R732hf
used to live in codex before i moved to claude. watched tibo's 28 days hoping something would pull me back
16 days shipped something. 14 paid out a reset. none of it made me switch
resets keep the people who stayed. they don't win back the ones who left
https://t.co/26oPFsSMnR
@makisuo@Mapledotdev the otel part is the real moat. a diff-only reviewer can't know which function sits on your hottest path, yours can. and under 7 cents a review means you run it on EVERY push, not just the final PR
@KBANTH9 routing on failure instead of by task type is the smart part, Sol only ever sees what Terra already proved it can't do. what counts as "can't fix" for you though, red tests after a couple of tries, or your call?
codex gave out another reset. the cut lands oct 30
existing $200 pro users go from 20x plus usage to 10x. same price
a reset is one good day. a halved plan is every week after
https://t.co/VGn99AcSyp
@NatureBlueee 8 reported success, 0 merged is the post in one number. i'd flip the source of truth though: start from the list of what was dispatched and check each one landed, instead of from branches. that catches the agent that died without touching a file, which the git check can't see