Detailed Analysis
A developer has published an open-source Claude Code skill specifically designed to restructure Claude's conversational output for users with ADHD, available at github.com/ayghri/i-have-adhd and activated via the `/i-have-adhd` command. The skill enforces ten behavioral rules that prioritize directness and cognitive clarity: leading with action rather than context, numbering steps sequentially, restating the current state at each conversational turn, and eliminating both preamble phrases and closing pleasantries such as "Great question!" The implementation requires no hooks and is framed as accessible to anyone who prefers this communication style, regardless of whether they have a clinical ADHD diagnosis.
The theoretical grounding of the project distinguishes it from ad hoc prompt engineering. The developer explicitly bases the ruleset on *The Adult ADHD Tool Kit* by J. Russell Ramsay and Anthony Rostain, a clinically oriented self-management framework, and adapts those principles for the specific dynamics of large language model interaction. This represents a meaningful translation effort: ADHD cognitive support strategies — which typically address task initiation, working memory load, and attentional fragmentation — are reinterpreted here as constraints on LLM verbosity and structural unpredictability. The approach acknowledges that LLMs, by default, are trained to produce socially warm, verbose responses that can impose significant cognitive overhead on users who process information differently.
The project fits within a broader and accelerating trend of user-driven behavioral customization of AI assistants. Claude Code's skill and hook architecture has created a layer of programmable interaction design that community developers are actively exploiting to serve niche but underserved populations. Where official AI interfaces are optimized for median user preferences, open-source customization layers allow individuals to realign model behavior with their specific neurological and cognitive profiles. The ADHD framing here is particularly notable because it signals growing awareness that AI usability is not a monolithic problem — the same verbosity that feels reassuring to one user can be genuinely disabling for another.
The developer's invitation for community feedback on which rules "feel off" or are missing reflects the iterative, evidence-informed spirit of the project. It positions the skill not as a finished product but as a hypothesis about what ADHD-supportive LLM interaction looks like, open to refinement through lived experience. This participatory dynamic — where end users with direct stakes in the outcome shape the tool's evolution — mirrors broader open-source software norms but applies them to a domain, cognitive accessibility in AI, that has received comparatively little formal attention from major AI developers. As neurodiversity awareness continues to grow in technology communities, projects like this may serve as reference implementations that eventually influence how AI developers think about interaction design defaults for diverse cognitive profiles.
Read original article →