Terretta

Thank you for writing this yourself, it's informative and refreshing!

If an LLM wrote it, then it's the best simulation of non-LLM writing I've seen on an agent related post. They seem all to have gone through an LLM "edit" (cough) that strips the personality seen in your update.

Also, Flint is awesome.

Engineering's stereotypically low empathy for those who may not connect dots as readily is no problem if teams would make a habit of these simple mechanical compensations.

In my view, the current "value" is driven by the current artificial scarcity of quality web search results.

For now with not all agents controlled by advertising engines, and not all content yet designed to honeypot them, today's agents make good headway bush-whacking their way to finding the right information on which to base task work for you.

If the first page of SERPs for anything, and the extra time needed to suss out quality information to inform your work or task doesn't bother you, then this benefit isn't one you'll value.

Put another way:

Today's agents help "cut through" the same maze of value extraction that Silicon Valley (where market share means find a way to tell users your thing is "free"), adtech, and the content-as-ad-scaffolding industry have built over the past three decades. Why waste time wading through all that when your time is better spent building more of it? :-)

Consider https://soulver.app/

On the splash page, check the calculator 'tab', or the docs it links to: https://documentation.soulver.app/syntax-reference/general (see left hand nav for categories of calcs).

See what's new in v4 for an idea of higher level capabilities (these are just new ones): https://documentation.soulver.app/whats-new-in-soulver-4

> can't communicate well

Or, perhaps the person thinks (solves) in abstract concepts rather than linear words. Just because many people mouth words when reading or can't think without hearing words in their head or aloud, doesn't mean others should have to handicap their own efficacy of thought while working it out. Two conceptuals can communicate in a kind of ultra condensed shorthand liner verbal thinkers struggle to follow. Less can't communicate, more can't conceptualize squared off with can't verbalize.

Both groups should synthesize simply and clearly at the end, both should work to get in sync. And both ways of thinking have their uses in a team. Catch: conceptual thinkers can interview verbal thinkers, the other way around is more challenging.

While visiting not(USA) recently, found Xbox Cloud streaming no longer viable (lag fest) though last year travels (even over 5g hotspot) it was fine. Doesn't bode well for this streaming concept.

Looked for a real console and discovered all small shops and big box chains offer Nintendo and PS, none have Xbox. Speaking with managers revealed they or their suppliers "lost" or "ended" contracts to carry it.

Secondary market / used was dirt cheap – good condition consoles for less than a new controller: €35 for Xbox One S, €85 for Xbox Series S.

Smelled like a death spiral well underway.

Exactly. It's not the bricklayer's responsibility when mortar fails. It's the person who specified that mortar.

So I'd agree it isn't not a coder or developer's responsibility when they're given a bill of materials and told how to build, but it is an engineer's responsibility. Comp should recognize that as well.

The truth is in the url.

Also on iOS, using native content blocking and Wipr 2 w/ Wipr Extra extension. No ads.

https://kaylees.site/wipr2.html

https://apps.apple.com/us/app/wipr-2/id1662217862

// by solo dev making this for some 15+ years now, worth supporting even if u prefer e.g. 1Blocker or AdGuard or etc.

FWIW, a Nintendo Switch USB-C to HDMI cable with power input travels well for this, albeit a bit bulky.

This one puts the charger input at the HDMI end where the TV and presumably outlets are, so power comes to your video playing device over the USB-C at your device end.

https://www.walmart.com/ip/USB-C-HDMI-Cable-Adapter-Nintendo...

> many fewer realize that the other three copies of your iMessage data on Apple’s servers are just, uh, not encrypted

With Advanced Data Security enabled?

Of course, then there's the persons you're talking to. I figure they are one of your "other three", another is the iCloud copy (deduped from backups) ostensibly resolved by advanced data security, what's the third?

Brilliant example of leveraging conceptual compression through metaphor.

Most valuable paragraph in the piece:

“The model does more than expected. Each role's skill file starts with a few sentences framing the role… This lays the groundwork as the underlying LLM knows very well how a teaching hospital works, what a differential diagnosis is, why a handoff has a structure, and what it means to be a second opinion. When you tell an LLM a persona to embody, it does it very very well.”

This took OpenAI and Anthropic some time to grasp, arguably only landing in late 2025 or early 2026, when you see them start leaving out common sense instructions, and see system prompts dropping in size. They said this was thanks to newer models, but arguably it's from no longer fighting core concepts the training compressed in.

Exercise for the reader: what metaphor best captures your system?

• • •

Below the line feedback:

These kinds of lines are painful to even decide whether to read:

“In July we audited them. Twenty-five skill files, a little over 100,000 words. About 23,000 of those words, 23%, were flagged as removable without changing a single gate, command, template, or sentinel. The dominant pattern was the same rule restated in several places: six thousand words of it across seventy findings. One file was forty-seven percent redundant.”

By contrast, these kinds of lines are refreshing:

“While this merged rather recently, and we don’t have good data to point to its efficacy, early signs seem positive.”

Fascinating example of how unfortunate the mélange.

> This feels like throwing a sandwich in the trash rather than let a hungry person eat it.

More like: giving up on making artisanal sandwiches to leave on a counter for free, after realizing the surge in people grabbing them aren't hungry, they're running off to resell them in sandwich shops.

Given caching, blocking, userscripts, etc., most client things are have things built in that are effectively smart proxies or glorified layer 7 application gateways.

You can write your own client for arbitrary services. You can run your own client on your own hardware, and you can host your client on someone else's hardware. Your client can render the content however it likes. You can have someone else do the work to render it like you like it and run that, locally or remotely.

There are a lot of ways to get to an "apparatus" that's analogous to both nitter and to a proxy for nytimes that ensure consuming public comments by other users doesn't suck.

As far as I know, there's no law requiring billboards to be seen by all drivers using the public road it sits beside, nor requiring all web clients to render ads. As far as I know, in most of the world there's no law saying users can't buy DNS or proxy or application firewall or ad blocking services from a third party -- services that can run between you and host, on your router, on your device at various OS levels, or as extensions to your browser.

On the contrary, there seem to be increasing numbers of laws requiring servers to deliver their content to arbitrary clients even when a firm makes its own servers, content, and preferred client devices and wishes to "protect the user's experience" or whatever the firm's justification might be.

Given a decade or two of marketing the "public square", other people's public posts, intended for public consumption, and expected to reach the public, picking that content as worthy of banning "useragents" should be shaky ground. Should be an interesting battleground for digital rights.

It's enough to wonder if a reason to go after the nitters is to establish misfortunate legal precedent using a target that can't defend itself?

Again, they give you the tools for this. You can understand your usage by API, you can set threshold actions on it, you can switch things on and off, and applyable templates exist, all of which are under your control and "out of band" from the systems delivering traffic or compute to your paying customers.

Mistaken view.

I was a member of AWS customer advisory board for years. Two dozen of largest enterprises and unicorns and so on. Nobody wanted the billing department "inline" of systems controls.

AWS provides all the controls you need to flip the switch yourself, and provides the CDK / TF / etc. patterns for it. Plenty GitHub projects to host your own light switch service.

Don't relinquish the switch.

FTA:

> Nobody wants to wake up to an email sent at midnight warning about a budget limit and find that, while they slept, their rogue service had consumed several hundred (or several thousand) more dollars of usage.

Monitoring and a circuit breaker. If you are given the tools, make your own heuristic and flip the breaker yourself. Don't let someone else turn it off, and be hostage to their process failures for getting it back on. By self-selection, if you are a hard limit customer, you are not front of their service line.

> An argument against this is that businesses don’t want their hosted applications to start throwing errors because some budget was exceeded. I expect that most businesses and individuals would prefer errors to a surprise $10,000+ bill.

Based on a career serving the Fortune 50, CTO for trillion dollar bank, etc.: "No." Assuming actual business is being done by the machinery in question...

The lights must stay on.

Cost management cannot shut off the enterprise. Cascading costs, even before reputation, are incalculable.

Fwiw, this has been helpful for a couple decades now:

https://symless.com/synergy/features

Given that even a corp Mac can now (policy permitting) allow a personal AppleID along with the corporate login, it's easy enough to do the same with just native OS controls even with 2 logins. Something magical about opening an iPad next to a MacBook and pushing the cursor across ... or, alternatively, wirelessly (or USB-C) making the iPad an extended screen.

But Synergy remains great.

If free is your bag instead of supporting the nominal cost:

https://github.com/deskflow/deskflow

They explain this here:

https://symless.com/synergy/blog/how-to-use-synergy-for-free

heart recovery is one of several headline new features they mention on first run (and in app ref info) will calibrate over time

as often the case when Apple makes its move, their implementations -- while late compared to e.g. Garmin which has had them for years -- purport to be leading accuracy on these features among wearables

Very little ‘magic’ seems necessary for those, rather a little less human-in-the-loop team sports with our future, and a little more feedback loop science.

Self-selected ignorance and scrambling seem both correlated and causative.

That said, SOTA machines can't run an office snack bar yet, so it'll be a minute before abundances other than caution.

Love the idea…

Except the "one model" is too small for a 64GB Mac much less 128GB, sad since the Q3 is proven less competent.

Offering a Q boost (with no leave behinds) on first run would be a bump worth some vibe coding while.

Perhaps one wishes to remote KVM consoles.

Was posting "Now make it join Tailnets." to the comments then saw your remark.

TY for link.

Note that Mac Studio Ultra M3 (or upcoming M5) with 512GB is effective. You don't have to do this work fast, overnight is fine.

Unless trying to use it interactively and adversarially, in which case it's not fast enough plus would be why those of us without our own datacenters will get told we can't have nice things.

> you forgot about the deps… (… 700Mb)

On the contrary, the readme for this project, `litelm`, is contrasting with the bloat project, `litellm`. This project subty deboats an L.

The bloat (or featuritis) project has the 700mb and 100k+ LOC, this de-bloat has 2k LOC and specifies its only two deps ones most people want as deps (although httpx should be swapped to its successor).

So this response is off base, the 700Mb isn't forgotten, it's been dropped.

There it is:

"Congress gripped by AI panic after doomsday warnings"

https://www.axios.com/2026/09/11/congress-ai-anthropic-coxon...

> why do we get a post like this from openAI then immediately after get the same thing from Anthropic and vice versa like it’s a single entity deciding what the next topic is to relay to us simple folk

Not relaying to us simple folk. The message is for government simple folk, amplified and relayed by us simple folk as constituents.

The common element is lobbying, for regulatory capture, to help pull up the ladder.

They don't have to be colluding, they just have to hear the same things from the gov at the same time (as they would), then it goes in media waves beacuse journalists don't as easily get published for a story about one thing as they can if two or more examples make a pattern.

If what you're doing makes money, the last thing you want is to stop doing the thing making you money. If, instead, it costs you money, don't wait until a usage surge to consider how to handle that – set it up now.

You also don't need any latency in the loop or moving parts that could wrongly interfere with the reliability of making that money.

Fortunately, the last thing this is is "opaque". You have all the observability you need to do what you want to do with unexpected usage.

What AWS (and every CSP) gives you is all the instrumentation and out of band tools to implement your own angel of death, to API kill a runaway thing if that's what you want to do.

You can also find countless TF and CDK patterns for it if you aren't sure how.

But please, no, do not ask to make critical systems subject to the whims of some billing decision enforcer.

The "why did they cut me off" complaints here far outnumber the "runaway bill" complaints.