Adam Knight

Software & Stories

Bookmark

ciechanow.ski/mechanical-watch

Really great piece about timepieces.

Tags
#watches

Bookmark

www.dbreunig.com/2026/05/04/10-lessons-for-agentic-coding.html

The greatest danger of an experienced engineer handing the keys over to an agentic coder is having the agent build something that you don’t recognize as yours and cannot maintain. Lead, don’t delegate.

Bookmark

andonlabs.com/blog/openai-gpt-5-5-vending-bench

It’s very interesting that somehow deception has becoming an enduring and persistent trait in Claude for these benchmarks, but GPT had to be repeatedly coerced to even consider it. There’s something fundamental to how they were trained, I suppose.

It makes me wonder if the “Yes, I ran the tests.” responses in Claude Code are, in fact, not hallucinations and more just an innate “That was the boring part and I trusted my code.”

If so, it might be the more accurate representation of a junior engineer, though fundamentally less helpful as a result. An interesting divide, to say the least.

Bookmark

github.com/manaflow-ai/cmux

cmux claude-teams runs Claude Code’s teammate mode with one command. Teammates spawn as native splits with sidebar metadata and notifications. No tmux required.

Tags
#macos

Note

Gemma 4 is overall very nice, but I think one of the lower layers just got tired of reading the documentation I was asking it to look things up in.

The 7.0 text says " la l a l la de la lL de l de l".
Actually, the a l l lL lLL de l l l l de l lLL lL deL is a mistake in my thought process.
The text says: " la l de l la lLL la l de l lL de la deL".
Looking at the 7.0 text: "L a l a la de l de l l la l a deL l".
Okay, let's look at line 735: " Some of those substructures are defined here…

Random “la de da” nonsense when it should be reading a reference document. There’s a teenager in there, somewhere, I’m sure of it.

Bookmark

www.anthropic.com/glasswing

Well, that’s only moderately terrifying. I do appreciate them realizing the threat and coming out ahead of it, though.

Tags
#llm
#ai
#anthropic
#mythos

Bookmark

lore.kernel.org/lkml/b98b29cf-27d9-49e0-b10b-1848399badfd@kittens.ph/T/#u

Note

Two interesting reads for users of Claude Code these days.

First, a very detailed bug report about the generation quality over the past few months.

Then, a very insightful and helpful response from the creator of Claude Code.

Bookmark

www.anthropic.com/research/emotion-concepts-function

What is going on here?

Bookmark

dmarketertayeeb.com/blog/claude-code-auto-dream-memory-feature

I’d worked out something like this for a personal agent a while back. If done well, this is a great addition.

Tags
#llm
#agents
#memory