mceachen

I bought a cheap pack of 5 colors of 3m electrical tape. Each cable gets one or two pieces of tape:

Data:

White if usb2

Blue if it supports USB 3 data (because 5Gbps USB 3 ports are mostly blue)

Green if TB/USB4/40Gbps

Power:

None if low/no Power Delivery

Yellow for 45W PD

Red for 65W+ PD

Have your agent make a plugin marketplace -- mine is here (feel free to crib): https://github.com/photostructure/coding-skills

Be sure to increment unofficial plugin versions when you make edits: codex's auto-update works reasonably well, claude not so much, but when asked, both can fix their own config.

And like others have said, imho the skills that are incanted as macros are much more reliably useful. I use my technical project plan skill suite in 90% of my sessions via direct reference, and the stage -> cross-model second-opinion review is how I land all my commits.

The neurologists and neurotherapists for a family member who is a stroke survivor have been pretty consistent in saying that cardiovascular exercise and regular social engagement seem to be the most promising activities to stem cognitive decline.

"Use subagents as applicable. Use sonnet/luna for finding relevant code references"

This should have been the entirety of TFA.

Aldus Freehand 1.0 was released in 1988. Adobe Illustrator was first released in 1987.

Sol doesn't support larger contexts, and the prior fable context was ~300k, way under where you see substantive regression

I had two sessions this morning that prior fable and sol sessions were stuck on, where iterations just resulted in _different_ bugs. (One kind of tricky fe layout problem, the other was a backend refactoring that was complicated by trying to aggregate a couple prior sessions that crashed).

I summarized each into new fable 5.1 sessions, and both seem to have arrived at reasonable solutions that only need a few nits revised before they are commit worthy.

age of xitter won't make you feel better.

The general workload performance (browsing, text editing) is comparable for neo and m* MacBooks afaict.

Ghastly was macOS on an Intel MacBook Air.

"... that matters" is the new emdash

Model competence is an interaction of weights, system prompt, and harness.

I suspect most CFOs/CTOs fear the risk of undertaking a rewrite of your payment processor or merchant of record layers (which can be tricky to fully test even with sandbox/test environments) more than the percentage discount and opportunity cost of not shipping more features.

If you're referencing Fetch for mac (pre macos) or Transit by Panic (later renamed to Transmit), I was a fan (and I've been a terminal fanboy since the early 80s). Both of them started as FTP wrappers, and both provided convenience the cli lacked.

Claude is quite good at searching both its own and codex session transcripts, and if asked, can spawn tmux/konsole/whatever to resume those sessions to be interactive if needed.

My system prompt adds TPPs and TLDRs for task/feature planning, and shared memories/documentation management. I find them invaluable.

https://photostructure.com/coding/claude-code-tpp/

"Wdyt" is a prompt I've used for ages (originally to get around Claude "ultrathink"-related parsing bugs), and it's helpful alongside replan

https://photostructure.com/coding/claude-code-replan/

Please don't conflate a volunteer effort with no expected economic gain with a very well funded company (or fleet of companies).

With the CCP's highly successful track record with subsuming other markets, Occam's razor applies to why they're doing this.

All of those phrases I've heard actively used even a decade (or two) ago. (I actually had to read your comment twice because I thought you were saying always, not never!)

"Critical path" and "long pole in tent" didn't make it into the model training data, but those were certainly also in play incessantly.

But they're all reasonably useful descriptions for common things, so I'm not surprised.

I don't know if it's universal, but your non-dominant hand may form reflected letters delightfully easily if you air-write simultaneously with your dominant hand.

If you're really wanting to do perceptual hash based deduplication, use multiple, heterogeneous hash algorithms (phash, dct hash, mean hash, ...) as it is likely that a given hash algo will happily lossily match with very very different images--but if all hashes match, you're much less likely to have false positives.

I wrote up what I do here: https://photostructure.com/guide/what-do-you-mean-by-dedupli...

More than 3 channel displays have been attempted by several companies, but I suspect they fail due to:

1) trying to convince content makers to use new custom high-gamut hardware to capture the new spot colors

2) you'd need a full video content production pipeline that can render to that color space

3) finding enough people to care enough to pay the (substantial) premium for niche production numbers.

4) Most content just doesn't warrant high gamut unless it's narrated by David Attenborough.

So, you have both a chicken and egg problem, and not that big of a TAM to warrant the struggle.

Here's how to do it with more recent versions of Hugo:

https://photostructure.com/coding/hugo-markdown-output/

(It includes the grandparent's head link suggestion, but it's not just "change .html to .md" because I'm old skool and as a wee nerd was told that URLs ending in .html or .php or whatever we're frowned on, so the above link's markdown is available by appending /index.md )

It's worse: Google added and then dropped even experimental support of jxl in 2021-2022. Several years later they adopted the rust jxl library, but have kept it behind experimental flags.

"No nukes" was Sun Microsystems lawyers' liability reduction, not a political statement.

See https://www.lawinsider.com/clause/note-on-java-support/_2 for the specific verbiage and diffs over time.

If your engineer tells you that, you're going to have a bad time.

I think you're thinking this:

> If you are not embarrassed by the first version of your product, you've launched too late. -Reid Hoffman

Nope, current flagship models are very happy to make huge missteps across the whole development stack of design, planning, implementation, and testing -- but playing different models against each other can help catch more egregious issues.

This is a very recent model behavior change: for me, Opus 4.6, Gemini 3.1 Pro, and ChatGPT 5.4(ish) -- prior models and harnesses suffered much more from sycophancy.

(I still prompt some questions and reviews with "our intern suggested..." to allow models to judge the quality of the content apart from the messenger)

I've found this surprisingly effective. Higher "thinking levels" may result in more than one approach being considered, but you can also tell your LLM to do brainstorming explicitly: https://photostructure.com/coding/claude-code-replan/

I applied it to PhotoStructure more than a year ago—I think it was when I first saw this on HN?

My point is that there *is* such a thing as software that tries to respect their users. I'm not the only one.

Random crowd anecdata is still anecdata.

Since when did corporations care? Most seem to just pay their insurance premium for cyber liability and call it a day.