theletterf

Logged in to Archive.org profile, where you're supposed to manage "subscription". The site lets you donate recurrently even when unlogged, obviously.

They do, but require that an incident is raised claiming fraud, etc. I guess they want to avoid misuse of this feature.

Not if Google Pay has been used as the transaction agent. It's not a Google Play subscription.

Just did. Not possible in Spain.

Well, you can't if you paid using a debit card, like I did. Silly me.

It clones the skills repo if not present and relies on the git last commit as the "version".

Be careful: paying with Google Pay is extremely easy and the default setting is monthly recurrence. If you're not logged in, the only way to cancel (as far as I know) is by sending an email, which is incredibly precarious.

We keep the skills in a repo, where an agentic workflow runs biweekly to check if their content drifted compared to the docs and opens PRs if they did. The repo is also a Claude plugin. The biggest problem is keeping skills up to date across users, so I developed a small Go binary that takes care of that across harnesses.

Ah, a fellow thalassophobic. Yeah, no, I'm not going to swim in the vast ocean at night, thanks.

https://en.wikipedia.org/wiki/Thalassophobia

Docs engineer here. Nice to read about writing style: would you consider creating a writing benchmark at some point? I guess y'all are painfully aware of the load-bearing issues (pun intended).

Seeing all the questions in this thread, I strongly recommend Dachary Carey's post on the topic of agent-ready docs: https://dacharycarey.com/2026/02/18/agent-friendly-docs/

I think it's easier to serve those bits as Markdown to agents only through content negotiation.

Agreed on the first experiment, that's why the post says it hit the ceiling. The others are not saturated. I'd love to find a way of making the first experiment more discriminating.

Code doesn't explain best practices, architectural concerns, and so on. Best combo? Docs + code.

Indeed. Another thing worth exploring is how much the position of instructions matter; Claude claims to read docs as a whole unless they're very large, though it might be that it only grabs the first lines of a doc sometimes, in which case there is an "above the fold" effect.

I wonder if there's room for a Linux distro á la Microsoft Bob. :)

http://toastytech.com/guis/bob.html

On one hand I'd love to browse the site; on the other, my PTSD would prevent me from doing so.

I wonder if ads will be sort of elegant and contextual, like product placement in screenplays.

"I've completed the analysis. One moment...

_Takes a sip of Evian_

Ahhh. Now where were we? Ah yes. Here are the results:"

I wonder if these models, left running in an air-gapped environment, would end up developing honeypots for the first human to log in.

There's a certain etiquette to it. When someone asks me something and I use an LLM to research the answer (notice that I said 'research', not 'produce'), I frame the findings as something that I've investigated with, say, Claude. I also take the time to edit the response and read if myself at least twice, because I don't want a stochastic machine to take over my identity.

This is fantastic, but I so hoped it would come with an infographic of NY's underworld, slice of cake style. I recall seeing one somewhere, but my attempts at finding it again all failed.

The Windows 3.1 theme of my blog says Hi: https://passo.uno/win31/fine-tuning-docs-llm/

Ha! I didn't know the origin was Shannon. It makes sense now. Big fan of my namesake Bellard. :]

Try Runpod or similar services: you can fine-tune stuff for the price of a latte. Stanford's NLP course recommends them: https://cs336.stanford.edu/

I concur. Old docs are the good stuff: https://passo.uno/why-collect-read-old-computer-manuals/

Interesting! I think the advantage of style fine-tuning is that you might not have to provides that much context upfront. Also, it's kind of magical to have an LLM just do something out of the box. I'll compare my local fine-tuned models against the baseline with instructions and see how they fare.

Had to pick a year, and most of the material hovers around the mid-90s, the golden age of MS docs. And 1995 is THE Microsoft year. :)

Author here. Sorry, should have made that more visible!

Author here. Another trick, I would argue, is consistency, hence the focus on style.

I also wrote on what I think makes docs beautiful, by the way! https://passo.uno/what-makes-docs-beautiful/

I sometimes long for audio feedback when interacting with Claude, so I've made a Claude Code hook that makes Claude reply to negative feedback with sounds from a folder. You can use any sound you want.