Karl Seguin's Blog - A mix of coding and creative writing

A CLI tool for analyzing Claude Code/Codex CLI usage from local JSONL files. - ryoppippi/ccusage

Close to three years ago I wrote about using Oddly Specific Objects' alternate "motherboard" to modify a classic Casio F-91W watch:  Pimping...

Download YouTube No Translation for Firefox. Keeps YouTube content in its original language (Titles, Audio Tracks, Denoscriptions...)

Brazil's Supreme Court issued search warrants and restraining orders against Bolsonaro on Friday, banning him from contacting foreign officials over allegations he courted the interference of U.S....

1/N I’m excited to share that our latest @OpenAI experimental reasoning LLM has achieved a longstanding grand challenge in AI: gold medal-level performance on the world’s most prestigious math competition—the International Math Olympiad (IMO).

Why I avoid using LLMs as a publisher and writer

Felix Baumgartner dies at 56 in paragliding crash near Italy’s Adriatic coast. Known for 2012 stratosphere jump.

gpt-5-reasoning-alpha-2025-07-13

h/t @swishfever

The method was pioneered by UK scientists to overcome devastating, often fatal inherited diseases.

Linux users who have Secure Boot enabled on their systems knowingly or unknowingly rely on a ke [...]

Ring reverses its plan to stop sharing video with law enforcement.

We are currently investigating a potential supply chain security incident involving the eslint-config-prettier npm package. This widely-used package, which helps developers maintain consistent code formatting by turning off ESLint rules that conflict with…

An important PSA.

The future of software development might just be jazz. Everyone improvising. Nobody following the sheet music.

Feeding slop is an act of war

The Matrix.org network has great potential, but after years of dealing with glitches, slow performance, poor UX, and one too many failures, I’m done with it.

We need to augment our command line tools and design APIs so they can be better used by LLM Agents. The designs are inadequate for LLMs as they are now – especially if you're constrained by the tiny context windows available with local models.