'i-have-adhd' Skill Trends on HN, Cuts Coding-Agent Preamble

A GitHub skill called i-have-adhd, published by a developer using the handle ayghri, cracked the Hacker News front page with 324 points and 256 comments in about 12 hours by imposing ten rules on coding agents that force them to state the fix first and drop the throat-clearing. The repo describes itself as "A skill to stop your coding agent from burying the answer. ADHD-friendly output."

The rules are blunt. "Lead with the next action." "End with one concrete next step." "Suppress tangents." "Cap lists at 5 items." The tenth rule is the tone-setter: "No preamble. No recap. No closers." The README pairs a before/after example, replacing a response that opens with "Great question! Let me think about this..." with "Run `npm install jsonwebtoken@latest`, then edit `src/auth.ts:42`." The skill installs across Claude Code, Cursor, and Gemini, with configs for Kimi, Qwen, and OpenCode. License is MIT.

The comment thread reads as a running airing of grievances about Claude in particular. "Claude (in particular) is a terrible writer," wrote user jp57. "There's a whole cottage industry of skills and CLAUDE.md instructions trying to push it toward writing better." Commenter ryandrake singled out a specific Claudism where the model narrates what it didn't do: "I edited this.py and that.py but I _did not_ edit README.md and I did not commit."

The repo is now sitting at 30.7k stars and 1.9k forks. It lands as another entry in our coding-tools beat, our 152nd coding-tools story in the last 90 days, and reads as a user-run patch on frontier-model prose. The repo's own tagline: "Action first. Steps numbered. No 'Hope this helps!'"