Haiku 4.5

1. Update: My viral consumer-rights AI game just went B2B - built with Claude Code + Opus 4.7 (www.reddit.com)

   +21 7w haikuopusclaude-code

   A few months ago I posted a small game here where you argue with an AI shop that won't refund you. It went viral and changed where this is headed.

2. F-Bombs Per Thousand Prompts (fpk): I measured my frustration across 44,212 Claude Code logs (www.reddit.com)

   +15 7w haikuopusanthropic+1

   Posted a writeup on a metric I've been tracking across 5 months of my Claude Code logs: fpk = f-bombs per thousand prompts. Frivolous-sounding, surprisingly real signal of developer friction.

3. Claude Design guidelines/benchmarks on model usage? (www.reddit.com)

   +12 7w haikuopus

   Using Claude Design for an app initially for web, later for mobile. On the max plan, which works well for the coding agents but Claude AI with Opus 4.7 can consume weekly usage in day 1 (currently Claude Design has it's separate usage).

4. [RELEASE] - coding agents can now talk! (www.reddit.com)

   3 7w haikucodexclaude-code

   Quick context: I use Claude Code and Codex daily and noticed I was spending half my "agent is working" time just sitting there watching the screen. I was like, what if Claude or Codex can just narrate its process back to me, so I know what…

← all threads