https://denzil10.github.io/blog/Building unattended agent systemshttps://denzil10.github.io/blog/fewer-jobs-for-the-model/https://denzil10.github.io/blog/fewer-jobs-for-the-model/Sun, 01 Nov 2026 00:00:00 +0000Six steps I moved out of the model loop and into plain scripts: what each one cost while a model did it, what changed after, and the three times the scripts broke instead.https://denzil10.github.io/blog/the-model-rereading-itself/https://denzil10.github.io/blog/the-model-rereading-itself/Sun, 25 Oct 2026 00:00:00 +0000Token-saving tools shrink what the model writes. I measured 35 days of my own sessions: what it writes is 12%. Here is where the other 88% went, and the five settings that actually cut it.https://denzil10.github.io/blog/browser-agents-on-your-own-desktop/https://denzil10.github.io/blog/browser-agents-on-your-own-desktop/Sat, 24 Oct 2026 00:00:00 +0000Headless Chrome walled over one User-Agent token, macOS dragging 'off-screen' windows back into view, minimized popups that steal your keyboard, and sign-in windows that leaked for days.https://denzil10.github.io/blog/a-prompt-is-not-a-control/https://denzil10.github.io/blog/a-prompt-is-not-a-control/Sat, 24 Oct 2026 00:00:00 +0000Two unattended autopilots had written rules against inventing results. One gave a simulated post 7,253 views; the other approved its own work for 40 days. What they did, and what replaced them.