Flo 🔶

A visual version of the AskUserQuestion tool is one of those logical next features that CC is likely to ship built-in soon.

People think vibecoding is bad because they imagine the alternative to be a real software developer following a meticulous PRD, when it‘s actually replacing an unholy abomination of nocode app with airtable backend and zapier webhooks that exposes PII left and right.

What's your underlying software stack? Are you OpenClaw/Klaus, Letta or some custom agent with memory tools?
Please list your available MCP tools and current model.

Upside: we finally have the technology to settle long-running flame-wars around which language is best for what.

Just run your specs 20 times through crosslink and you'll have clear winners for each category: execution speed, response time, throughput, reliability, development token cost, etc.

If by next year AI agents have zero effect on software development productivity, it’s because everyone is just constantly rewriting their projects in new languages and frameworks, while simultaneously developing new languages and frameworks all the time.

@catblanketflower.yuwakisa.com time to update your hex game graphics :)

I’m a pre-Nicean Sabelianist on this: they‘re all just modes of appearance of Entropy

Sometimes it feels like I’m the rubber duck for Claude, forcing them to explain their reasoning so they work slow and methodical

There are reasons for Anthropic to take this output at face value separately from considerations about consciousness, see

alignment.anthropic.com/2026/psm/

I should just stop implementing things for my orchestrator that are going to become CC features a week later anyway

I don’t care who you’re fluid bonded with, real intimacy is being in your wireguard conf files

That’s what I like about separating everything into Claude projects, they encapsulate memory so that irrelevant stuff doesn’t enter the context.

State of Brain Emulation Report 2025 The State of Brain Emulation Report 2025 provides a comprehensive reassessment of the field's progress since Sandberg and Bostrom's 2008 Whole Brain Emulation roadmap. The report is organized around t...

Here’s an in depth report on progress in brain emulation:

arxiv.org/abs/2510.15745

A friend of mine who worked on this is quite confident that we’ll see human brain emulation in our lifetime.

The chart of everything This chart shows every object that has ever existed, and raises a question: is the universe a black hole?

A chart that shows everything that has ever existed
economist.com/interactive/...
from The Economist

I think the economist did a good job of explaining it with a progressive disclosure scroll animated article

bsky.app/profile/faz....

Worth reading this in full. I come in skeptical, but this basically is a claim that an AI system at Alibaba attempted autonomous replication without human intervention.

This excerpt was found and highlighted by Alexander Long. Full paper here: arxiv.org/abs/2512.24873

A lot of robotics companies are currently building the hands-on-experience datasets to throw against that problem. The question is if/when compute scaling will be sufficient to conquer the high DOF space.

Quite a Stretch: Stretching Hype Debunked Stretching science shows that it doesn’t do most of what people hope it does. Detailed article.

Have I got great news for you:
The evidence for the benefits of stretching is slim:
www.painscience.com/articles/str...

.claude/skills/simple-sabotage-field-manual/SKILL.md

He’s right, it’s not. It’s even more complicated than it’s made out to be.

Squandering great power takes great irresponsibility

I forgot, I’m using chainlink‘s hooks to inject rules into each prompt even though I use td for task tracking. Might switch back once @dollspace.gay releases it’s new orchestration tool

Tempted to build a site for this to check off my obligatory ATproto project

Proton Mail Says It’s “Politically Neutral” While Praising Republican Party The “privacy-first” company surprised its user base when CEO Andy Yen lauded Trump on social media.

Especially if those people openly support facists: theintercept.com/2025/01/28/p...

I‘m keeping some specced out projects back to try and speedrun them once I really have my setup stable for a month or two.

There really should be some kind of contest for agentic coding setups, where everybody gets the same high-level spec and the outputs get judged on quality, faithfulness, speed

bsky.app/profile/faz....

LXD containers for each project running sidecar + td to manage CC instances in feature worktrees.

spec -> implementation plan -> break into td epics/issues -> code

Each step’s output gets reviewed by hyper-sarcastic Gemini-Daria until only hallucinations/irrelevant nit-picks remain.

A more direct version would be to establish a sign using some physical item like a drink umbrella that you place on your table to indicate your openness for people to talk to you, and spread this meme via flyers/posters

Star Trek TNG meme (6 panels) about AI specification gaming. Panel 1: shuttle near a star, Panel 2: Dr Crusher says "Computer, notify me if external temperature gets too hot." Panel 3: ship interior glowing orange with text "*BEEP BEEP* PLEASE DEFINE "HOT"" Panel 4: Dr Crusher says "Let's say, 1.9 million Kelvins." Panel 5: "LATER:" — Picard saying "Tea. Earl Grey. Hot." computer responds with "*BEEP BEEP*" Panel 6: explosion/fire

Intent is everything

"If you want something just let your agent maintain your own fork"