jrecyclebin

Oh nice! The hook installation response is clever.

My biggest frustration with MCP at the moment is that it has no built-in way of uploading and downloading files. Every argument has to go through the context.

Sure you can use curl along with it - but many MCP friendly harnesses don't have access to a tool like that. Wish they had added this before moving on to stuff like MCP apps.

This was my thought as well. I have a local model monitoring my finances and personal wiki - things I wouldn't want Claude to touch - and the Qwen 3.5 9b handles it all just perfectly.

I also needed a new device anyway - and having this much system memory to run virtual machines has been amazing.

Am paying subscriptions as well tho lol.

At first blush, seems well thought out. README is quite thorough and readable - really appreciate that you've worked on shipping on Mac. Don't know much about txiki.js - will have to play with it.

Good luck!

Was always my first tweak to a new machine - such an obvious change. Moving to a split keyboard gave me three keys at each thumb - the same kind of realization. We have two thumbs that most keyboards map to a single space bar. The thumb is a really solid finger for modifiers.

Another nice thing about a custom keyboard is not needing to map the keys - it's in the firmware.

Was always mystified about custom keyboards - until I got the idea to do an emoji layer: https://poggers.institute/@j/the-emoji-layer/

Also: swapping colon/semicolon, using fn+hjkl for arrow keys and having Enter at my right thumb.

Once you open that door.

Not a bad idea - however

> Caveman only affects output tokens — thinking/reasoning tokens are untouched.

The problem is the thinking. But could help to tune my system prompt for Kimi.

I absolutely love Kimi's personality - some of the things it says are so out there! And it's been great for very focused, iterative work.

Its weakness is that it seems to yak on-and-on when it needs to plan out something big or read through and make sense of how to use a niche piece of a complex library. To the point where it can fill up its 256k window - and rack up a build. (No cache.) I have had better experience with GLM 5.1 in those cases.

Anyone out there relate?

Well, yeah, it's all subjective - and actually quite tenuous - so you won't know good and bad until you actually make the call on it. Maybe you've had the experience even of coming around on some music you previously thought was bad.

Or like: one time I listened to a bunch of new music I had dug up and wasn't sure there was anything I liked. Two days later, I had a song in my head. Turned out to be one of the ones I had listened to. But I had to listen to everything all over again to find it! ദി(ㅠ﹏ㅠ) Glad I did - there were other gems in there.

Anyway, great quote.

Found a favorite band through a similar technique: pile of CDs given from a friend who worked at a music store and no one wanted them.

You have to be willing to sift through junk. Which I think is hard for many to accept. However, the algorithms are often giving you junk anyway. Kind of no way around it.

This is not true though. My two favorite bands from the past year were poorly-attended shows that I stumbled into. You can still seek out good underground, obscure artists - you just have to look for them.

Not trying to be elitist - like what you like. I just really feel like little artists need the support. Plus, it feels like there is a bit more satisfying agency and fate in looking for new things rather than being fed them.

Well, except that, in this case, Copilot really is for entertainment purposes only.

> There is no need for determinism to guarantee the job will be done identically every time if we only plan to do it once.

So can't you just save the conversation transcript and replay it with the tools? Seems a lot more efficient that regenerating the whole thing. And, also, no risk of branching when a tool reply is slightly different. (Of course, errors can occur on subsequent runs.)

Agent can still "forgot password" on many accounts. Or magic link.

Digitize and burn.

Another approach to journal writing is basically the opposite: rather than treating it like a task to fill with very rigid requirements - find a notebook and pen that you'll enjoy spending time with. An easy start is a Midori Ruled A5 (very simple, lay flat notebook) and a Uniball Zento Signature (the most hyped pen in the world right now) and treat them basically like little friends you spend time with. Writing only two sentences is denying yourself quality time writing and reflecting at a leisurely pace if you really come to enjoy it.

I'd also think you're more likely to read back if writing time is a fond memory.

Kind of ironic - the moat is money...

At the same time, I see the appeal. I feel like 10% of the comments I read lately are "is this an AI response?" - would be nice to be free of that. Probably not possible tho.

The weakest part is the last one - and it's a big one. Personalsit.es is just a flat single-page directory (of thumbnails, even, not content - so the emphasis is design.) To be part of the conversation, you'd list there and hope someone comes along. Compare with Reddit where you start commenting and you're close-to-an-equal with every other comment.

Webmentions do get you there - because it's a commenting system. But for finding the center of a community, it seems like you're still reliant on Bluesky or Mastodon or something. (Which doesn't "destroy all websites.") Love the sentiment ofc.

Great advertising for vidalias. I simply have to try one now.

Skill descriptions get dumped in your system prompt - just like MCP tool definitions and agent descriptions before them. The more you have, the more the LLM will be unable to focus on any one piece of it. You don't want a bunch of irrelevant junk in there every time you prompt it.

Skills are nice because they offload all the detailed prompts to files that the LLM can ask for. It's getting even better with Anthropic's recent switchboard operator (tool search tool) that doesn't clutter the system prompt but tries to cut the tool list down to those the LLM will need.

While I totally agree with you, I also can see a world where we just throw a ton of calls in the MCP and then wrap it in a subagent that has a short description listing every verb it has access to.

Yep, those are the companies that would go away.

Something of a logical leap here: if LLMs aren't capable of replacing workers and it's all lies, then what company is going to engage in mass layoffs without seeing results first? Sounds like companies that deserve to go away.

Lots of gold in this article. It's like discovering a basket of cheat codes. This will age well.

Great links, BAML is a crazy rabbithole and just found myself nodding along to frequent /compact. These tips are hard-earned and very generously given. Anyone here can take it or leave it. I have theft on my mind, personally. (ʃƪ¬‿¬)

Ok well - appreciate the further details.

Idk the punishment just doesn't match the crime. Can't they just confiscate the computer? Or pressure the ISP to cancel his account? Tbh I get that the Feds are going route around and through anything that stands in their way.

Instead we're left up to state thuggery.

I'm already not going back to the way things were before LLMs. This is fortunately not a technology where you have to go all-in. Having it generate tests and classes, solve painful typing errors and help me brainstorm interfaces is already life-changing.

One of the main things I put in my instructions is "hey I'm a solo dev and it's just you and me working on this stuff, so I'm looking for all responses to be concise." I think it helps to give your situation so that the output can be in the proximity of "solo dev" content - which is going to be more concise and practical by nature.

Kind of like telling it to generate Ghibli pics. These things are best at imitation.