Two Unrelated ADHD Agent Tools Trend for Opposite Reasons

ayghri/i-have-adhd sits at 16,022 GitHub stars as of August 3, 2026, up 5,225 in the week before that date, per its repository. It's one of two unrelated agent tools trending under the same name: ADHD. Coding agents that reason step by step tend to bury the actual answer under paragraphs of thinking, and both tools exist because someone got tired of digging for it.

ayghri's project, a Python tool under an MIT license, reshapes agent output so the answer stays scannable instead of buried in verbosity.

UditAkhourii/adhd takes the opposite route. The TypeScript project, also MIT-licensed, reached 3,141 stars as of August 3, 2026, up 746 in the week before that date, per GitHub. Built on the Claude and Codex Agent SDK, it runs tree-of-thought with pruning: fanning out parallel divergent thoughts under different cognitive frames, scoring them, pruning the dead ends, and deepening whatever survives.

So one tool shrinks what the agent shows you. The other expands what the agent thinks before it shows you anything, then prunes the expansion back down. Two different mechanisms, same name, same underlying complaint: agents burying the real answer in too much reasoning.

Two teams landed on the same disorder name independently and still built opposite fixes. That's the signal worth watching: expect more single-purpose output-shaping wrappers like these before Anthropic or OpenAI build the fix into the base model instead of leaving it to the ecosystem.