Join 📚 Josh Beckman's Highlights
A batch of the best highlights from what Josh's read, .
Armin theorizes that this is because more recent Anthropic models have been specifically trained (presumably via Reinforcement Learning) to better use the edit tools that are baked into Claude Code. This has the unfortunate effect that other coding harnesses, such as Pi, may find that their own custom edit tools are more likely to be used incorrectly.
Better Models: Worse Tools
Simon Willison's Weblog
The signs of whether a cutting has rooted or not is to look for the new growth. If it shows new growth - lush like this - then there's every possibility that it has rooted.
Summer Prune and Spring Follow Up
Herons Bonsai
It’s often very useful to provide inline performance information in a query response, for example when debugging a slow query. The GraphQL specification allows servers to include additional information as part of the response under an extensions key. This is incredibly useful for metadata like tracing information (e.g. in Apollo)
Production Ready GraphQL
Marc-Andre Giroux
...catch up on these, and many more highlights
Readwise