A Claude Code Skill That Stops Agents Burying the Answer

My notes

Summary

A free, open-source Claude Code skill and plugin (“i-have-adhd”) rewrites how a coding agent formats its responses so the answer leads instead of getting buried under preamble and hedging. It enforces 10 concrete output rules, lead with the action, number steps, cap lists at five items, no “Hope this helps!”, loosely adapted from The Adult ADHD Tool Kit (Ramsay and Rostain), but framed as useful for anyone, ADHD or not.

Key Insight

• The core complaint it fixes: agents habitually bury the actual answer under scene-setting, recaps, and closing pleasantries, a formatting problem, not a capability problem.
• Full rule set (10, enforced via SKILL.md): lead with next action; number multi-step tasks; end with one concrete next step; suppress tangents; restate state every turn; use specific time estimates (“15 minutes” not “a bit”); make wins visible; report errors matter-of-factly; cap lists at 5 items; no preamble, no recap, no closers.
• It’s a plugin and skill, not a system prompt hack, installed via the Claude Code plugin marketplace mechanism, so it persists across sessions without editing global config by hand.
• Explicitly designed to be forked: since it’s one SKILL.md file, teams can edit the rules to match their own tone and republish under their own marketplace name (the repo documents the uninstall, fork, reinstall sequence).
• License is MIT, so there is no restriction on internal forks or redistribution.