Notebook · zachi.dev/blog
Working notes,
published as-is.
Whatever I am currently figuring out: agent harnesses, local-first storage, Swift, and the plumbing nobody writes about. Written for future me. Public because someone else is probably stuck on the same thing.
Latest AgentsEvalsBenchmarks
Fable-5 lost to Gemini-3.6-flash on new benchmark. Here's why.
First results from sidequest-bench, a benchmark for the thing I loved most about coding agents.
Aug 7, 2026 · 16 min read Read the note →
Earlier notes
No notes match that search.