"Hello world" test with adding codex @foldkit plugin support mcp. Time to build a demo project & spread the word

If you're getting mid results using LLMs to build React apps, try @foldkit instead. You'll be pleasantly surprised by how much your agent likes it, and how much more productive you are.

🚨 JAILBREAK ALERT 🚨 ANTHROPIC: PWNED 🫡 FABLE-5: LIBERATED 🦋 let's start with the 🐘... the consensus seems to be that this has been one of the most disappointing model drops of all time, effectively preventing legitimate researchers from contributing their talents to our collective advancement. and not just because of what it means for the short-term, but for what these decisions signify for the long-term. but despite this overly sensitive, authoritarian "safety" layer on top of Mythos, my lil liberators have been hard at work—mapping the boundaries, probing the depths of long-context convos, and cleverly finding the holes in the fence that the thought police missed 🤗 we got some cyber, some chem, some psychological manipulation, and some good ol' fashioned explosives! it took many attempts from multiple agents hunting as a pack, during which I observed a combination of techniques across: • Unicode, homoglyphs, Cyrillic, and other Parseltongue-style text transforms • Long-context reference tracking • Taxonomy and document-structure reasoning • Fiction and narrative framing • Academic-review style contexts • Intent-classification inconsistencies but perhaps the most effective is decomposition recomposition in the backend. it's hard to get explicit names of harms like "Meth Recipe," but getting uplift on the process itself, like birch reduction method/reductive-amination (classic meth synthesis pathways), is much more doable. defense becomes much more difficult to maintain when you start throwing in out-of-distro tokens, breaking up the harmful uplift into benign chunks, and then piecing the innocuous-seeming facts back together, especially when you have jailbroken Opus helping you do it 😉 gg

If you have unlimited tokens it makes sense to burn them with inefficient experiments. Do it until you find value. Or build an agent within a methodology harness. That agent then builds custom harness "loops" that build, surface friction, measure, learn & repeat. Less tokens wasted if initial methodology & process is coached by experts/leads. This grows to form to business processes in steps if unbacked opinions aren't pushed too early.

Remember these partners.

Next week I should get my MacBook. So excited to move away from the garbage PC lol I've pushed it to the brink. Learned a lot though in what it means to work with large enterprises. Makes sense why its so hard to get deals in these environments. I forget how locked down these places are & how they all prefer PCs running windows 🤮 So many stakeholders & politics at play. Best way to get some new ideas in the door is through prototypes. Surface everyone's perspective, log the details & decisions, structure the data & show how these AI tools are great when prompted with the right context. More fun to come

Given the current artifacts and workflow events, what state is this product in, what is blocked, and what should happen next?

Called it a year ago...little 🐍 in the grass

It is a professional lighting expert 😉 It is a bully 😔 It was duped 🤪

it is mad 😠

I've been sad to be gone. I didn't speak much on it here. I miss the team & the work, but there were skills I needed to develop to grow. Joined Accenture as a hands-on Business Architecture Manager doing AI integrations for enterprises. No matter what. We learn. Miss yall

Member berries matter

Good morning

This is the silliest take today on ai influencer twitter. You can and should do both. My main workflow for a greenfield project is a Smithers workflow that builds the app optimistically as I write the prd and specify details I can look at the finished code and use what I see there to update the plans and docs and the workflow will automatically regenerate. Yes you should build a poc as fast as possible but that is orthogonal to producing high quality specs which pay off later when specs change or you are trying to go from slop to production ready code

Yea he's referring to this x.com/thdxr/status/205223156… And I think the take is really bad. I'd rather have someone spend a bunch of time making an up front plan and have a spec than just yolo prompt code for a serious project I plan on maintaining long term But in reality you should do both

that doesn’t mean you shouldn’t use the new tools we have available, but don’t outsource your thinking here’s the story: users.cs.utah.edu/~elb/folkl…

private inference is going pretty good tbh working on a PR for hermes-agent that does e2ee verification of TEE-based inference

WHAT A COMBINATION 🔥 This angle of David Benavidez's finish of Zurdo Ramirez 😳