Run Parallel Divergent Ideation With ADHD
Use ADHD to fan out isolated reasoning branches for open-ended coding-agent decisions, then score, prune, and deepen the strongest options.
npx skills add agentskillexchange/skills --skill run-parallel-divergent-ideation-with-adhd
npx skills add UditAkhourii/adhd; use -g for global install or -a to target specific agents. Restart the target agent, then invoke explicitly with /adhd or allow the skill to trigger on open-ended design, ideation, naming, and fuzzy-debugging prompts.ADHD is a coding-agent skill for expensive, open-ended decisions where a single chain of thought converges too early. The operator invokes it for architecture tradeoffs, fuzzy debugging, naming, API design, strategy, or prompts like “give me a few ways to…”. The workflow fans out isolated branches under different cognitive frames, scores and clusters the results, prunes weak ideas and traps, then deepens the strongest survivors.
What this skill actually does
Use this instead of normal agent prompting when the cost of the obvious answer being wrong is high and breadth matters more than speed. Do not use it for syntax lookups, routine bugs with a known root cause, or requests that need a quick canonical answer.
The scope boundary is the ADHD skill and CLI workflow: it is not a generic tree-of-thought library listing, and it is not a broad agent framework. The approved use is deliberate divergent ideation inside coding-agent sessions, with an explicit pre-flight gate because it costs more model calls than a direct answer.