TL;DR

Opus 4.7 got terse on me. Anthropic's release notes confirm it: more direct tone, less validation-forward phrasing, response length now calibrates to task complexity, more literal instruction following. Side effect: harder to follow on complex work.

Fix: a custom output style (thorough-explainer.md) that tells Claude to spell out the reasoning, define terms inline, stop making me connect the dots. Drop in ~/.claude/output-styles/, activate via /config.

Also built evals to catch this drift on future model releases.

Realization

I've been using Claude and Claude Code for a while now, and I've sat through several model updates without much friction. Opus 4.7 was different. Over the last few weeks I'd noticed I was having to work harder to digest what Claude Code was telling me. Not the code, but the analysis, the progress reports, the wrap-ups at the end of a task. They'd gotten terse. The harder the project, the harder it got to follow. So naturally, I asked Claude for help.

Continue reading →