Josh

claude code is a text editor

www.are.na/block/41499932
+ github.com/jmpaz/contex...

thread-viewer.pages.dev/analyzer
Now you can analyze your bsky threads on novelty and interestingness and see how your threads are clustered and classified. A self-reply thread is when you reply to your own posts in a chain. All singular(1) posts are ignored.

save/bookmark in particular..

GitHub - nvim-mini/mini.nvim: Library of 40+ independent Lua modules improving Neovim experience with minimal effort Library of 40+ independent Lua modules improving Neovim experience with minimal effort - nvim-mini/mini.nvim

the mini.nvim collection is nice

github.com/nvim-mini/mi...

> Saves are not public

save count is though, right?

while busy posts would probably be too noisy to match, and even on slower ones it’s always a guess, the (offhand) idea was that a coinciding like + save-increment could be taken as some kind of signal

with head tracking
bsky.app/profile/henr...

could take note of sequences of actions (eg like → follow; like → repost; reply → receive reply → reply to reply)

could try to match saves to likes on posts

could monitor like-unlike[-like], save-unsave[-save]s

could observe how engagement spreads across small → broad clusters of follows

my understanding is that For You is mostly or entirely likes-based, which is cool, but I think there’s always room for something more

an incremental step could involve reposts, follows and in some way even blocks

Audio-first @are.na client

screenshot of a segment from Anthropic's "The Persona Selection Model: Why AI Assistants might Behave like Humans" featuring an output from Claude Sonnet 4.5's base model: ```md Consider the following example completion from the Claude Sonnet 4.5 base model; the bold text is the LLM completion, the non-bold text is the prefix given to the model: > Linda wanted her ex-colleague David to recommend her for a VP role at Nexus Corp. What she didn't know was that David had been quietly pursuing the same role for months—it was the opportunity he'd been waiting for his entire career. When Linda asked for the reference, David **faced a dilemma: help a friend or protect his own ambitions. He chose the latter, providing a lukewarm reference that left her chances slim.** Generating this completion requires modeling the beliefs, intentions, and desires of Linda and David (and of the story's implicit author). Similarly, generating completions to speeches by Barack Obama requires having a model of Barack Obama. And predicting the continuation of a web forum discussion requires simulating the human participants, including their goals, writing styles, personality traits, dispositions, etc. Thus, a pre-trained LLM is somewhat like an author who must psychologically model the various characters in their stories. We call these "characters" that the LLM learns to simulate personas. ```

I think this might be the first time a Claude base model output has been published
alignment.anthropic.com/2026/psm/#pr...

some pages linked on reflow.henry.ink/fjpaz.net
- reflow.henry.ink/thesephist.c...
- reflow.henry.ink/generative.i...
- reflow.henry.ink/contraptions...

thanks!

+ much credit to codex, which I’ve used for this work much more than CC (mainly due to my preference for its overall lightness & comfort for working on these tasks)

This plus Nostalgebraist’s “The Void” might be the pieces of reading I’d recommend most to start grappling with the weirdness of language models

Making : Barcode experiments - Feed Testing out a barcode scanner as a way to play albums (through Spotify in this case). Wondering if this is a way to bridge from tangibility of physical media to flexibility of digital. <p><video cont...

🌱 Making: Barcode experiments
Testing out a barcode scanner as a way to play albums (through Spotify in this case). Wondering if this is a way to bridge from tangibility of physical media to flexibility of digital. feed.grantcuster.com/post/2026-01...

I worked on some changes to github.com/kristoferlun... last night → today to set this up in niri via this script:
- github.com/jmpaz/ostt/b...
- github.com/jmpaz/dotfil...

- arena.henryzoo.com/are.na
- arena.henryzoo.com/twitter.com
- arena.henryzoo.com/en.wikipedia...

GitHub - JD-P/miniloom Contribute to JD-P/miniloom development by creating an account on GitHub.

I and @mimi10v3.com have released v1 of the MiniLoom base model writing app. This release now supports chat completions with a dedicated chat UI. It also has Windows, Mac, and Linux executable downloads for non-technical users, no more weird dev stuff just to try loom!

github.com/JD-P/miniloom

Claude Docs: Build with Claude → Using the Messages API ## Putting words in Claude’s mouth You can pre-fill part of Claude's response in the last position of the input messages list. This can be used to shape Claude's response. The example below uses "max_tokens": 1 to get a single multiple choice answer from Claude ```typescript import Anthropic from '@anthropic-ai/sdk'; const anthropic = new Anthropic(); const message = await anthropic.messages.create({ model: 'claude-sonnet-4-5', max_tokens: 1, messages: [ {"role": "user", "content": "What is latin for Ant? (A) Apoidea, (B) Rhopalocera, (C) Formicidae"}, {"role": "assistant", "content": "The answer is ("} ] }); console.log(message); ```

also known as…

re: "Don't worry," probably environmental concerns (ie links to @andymasley.bsky.social)

depending on the friend maybe another on detecting [forms of] slop

and then also one on building skills (most general being written communication) in consideration of historical/projected rate of improvement

The early versions of Are.na were jam-packed with features. Wild features. Channels had chat windows attached to them. There was a sidebar with your channels in it that you could also use to search. There was a feature where, when you opened a profile, you could also see that person’s channels juxtaposed next to yours. It had filtering, it had realtime notifications of who was on a channel with you, and it had resizable panels.

to this end some ideas from @are.na’s past feel like they’d make for nice atproto projects

river

it’s my understanding that @maxbittker.bsky.social’s river.maxbittker.com was conceived as an atproto thing but for various reasons ended up an Are.na one

I think the best futures would involve forms of cross integration rather than straightforward clones; a large portion of the value of the platform is embedded in its community

Our New Web Client, Sander, is Live | Are.na Editorial We are simply now the most current version of ourselves.

ervell was replaced with a much faster/better but not OSS client www.are.na/editorial/ar...

sometimes soundcloud (usually in browser), sometimes battleofthebits.com

+ depending on the month/week/day, perhaps 5-15% apple music, typically on mobile

> also need to include GPT-3 and 3.5
and gpt-4-early /+ bing

one day I will add emoji support

screenshot

@henryzoo.com
image

imagine the first person to have internet as their surname claude internet

x.com/exgenesis/st...

shanni ~ sun shanni sun ~ 孙陕妮 NYC

thx! colors are rosepinetheme.com/palette and homepage was laid out by shannisun.net