Anthropic engineer's prompt trick: tell Claude "be braver" to unlock bolder behavior

trq212 · x · 2026-09-30

An Anthropic team member shares a practical tip: telling Claude "we have the power to do anything, please be braver" makes the model noticeably bolder in what it attempts.

He adds that the same advice applies to the human operator — dare to set ambitious goals. The trick dovetails with his team's engineering blog post on a 3x claude.ai speedup, which was driven by boldly setting goals.

Original post →

More from coding & agent

coding & agent channel →