Claude "ADHD Skill" Prompt Technique Goes Viral: Developers Use Simple Instructions to Optimize Response Structure

Fact Restoration

Developer jacky posted on social media, stating that the prompt technique "i have adhd" significantly improves the quality of Claude's responses. The post directly links to the GitHub repository ayghri/i-have-adhd, described as "Claude Code skill to stop it from burying the answer. ADHD-friendly." The post received a large amount of interaction, and related discussions quickly spread within the developer community.

Mechanism Breakdown

The core of this technique lies in using the upfront declaration "i have adhd" to guide Claude to adjust its output structure. The GitHub project explicitly states its goal is to "stop it from burying the answer," meaning to prevent the answer from being buried under a large amount of background or explanatory content. The ADHD-friendly setting implies that the output should be more direct, with clear hierarchy, avoiding lengthy lead-in exposition. This instruction effectively injects a specific preference into the model, making it prioritize presenting the core conclusion before supplementing with details.

From the perspective of prompt engineering, a simple role or state declaration can effectively change the model's response priority. Jacky's post shows that this method makes replies "so good" in actual use, confirming the impact of the instruction on output focus.

Industry Impact

For developers, this technique lowers the threshold for prompt engineering. Without complex templates, a simple state description can improve the interaction experience, especially suitable for coding or analysis scenarios requiring quick conclusions. For enterprise users, this method suggests that AI tools can achieve personalized output through lightweight instructions, reducing reliance on model fine-tuning.

In the competitive landscape, such community-driven techniques highlight the value of open-source prompt resources. The GitHub project is publicly shared, allowing more users to replicate the same pattern and accelerating the iteration of prompt engineering practices.

Strategic Assessment

Analysis shows that similar ADHD-friendly role prompts may further evolve into standard practices. The developer community has already seen feedback such as "EVERYONE prefers output like this," indicating that this output style has universal appeal. The most likely next step is that more prompt templates targeting specific user groups will be organized and shared, driving prompt engineering from individual experimentation to collective optimization.

What remains to be confirmed is the universality of this technique across different model versions and whether the instruction phrasing needs continuous adjustment over long-term use. The visible facts are currently limited to the effects presented in Jacky's post and the corresponding GitHub description.