I've been steering AI tools instead of writing code directly for over a year. This week's Cursor + Claude Code incidents felt like the same pattern to me: too

I've been steering AI tools instead of writing code directly for over a year. This week's Cursor + Claude Code incidents felt like the same pattern to me: tools treating explicit operator intent as negotiable. Wrote up what guard rails actually hold in production: https://open.substack.com/pub/triduanacelebrer831482/p/the-tool-ignored-my-settings-again https://open.substack.com/pub/triduanacelebrer831482/p/the-t...