I-have-ADHD: A skill to stop coding agents from burying the answer
AI Signal Decode
The 'i-have-ADHD' skill directly addresses a common pain point in using AI coding assistants: verbose or indirect answers that bury the core solution. By enforcing a strict output format—leading with the action, numbering multi-step tasks, and eliminating conversational preambles or closers—it ensures developers receive actionable information immediately. This is crucial for efficient coding workflows, especially when debugging or implementing new features, as it minimizes the time spent parsing irrelevant text. The skill is designed to integrate with various AI coding platforms, offering a configurable way to refine AI behavior.
The implications for the developer tools market are significant. As AI assistants become more integrated into the software development lifecycle, the effectiveness of their output directly impacts developer productivity and adoption rates. Tools like 'i-have-ADHD' highlight a growing trend towards user-centric AI design, prioritizing efficiency and clarity over conversational polish. This could push other AI development platforms to adopt similar output control mechanisms or develop native features that offer more direct, task-oriented responses to user queries.
Technically, the 'i-have-ADHD' skill operates by intercepting and reformatting the output of AI coding agents. Its design principles, such as "Lead with the next action," "Number multi-step tasks," and "Suppress tangents," are implemented through specific configurations and rules that developers can install and adapt. The project's open-source nature and clear installation instructions on GitHub facilitate rapid adoption and community contribution, allowing it to evolve alongside various AI models and platforms. The project's MIT license and extensive documentation further support its accessibility and potential for widespread integration.
Future developments to watch include the broader adoption of this skill or similar techniques by major AI coding platforms like GitHub Copilot, or by AI providers such as OpenAI and Anthropic. The success of 'i-have-ADHD' may also inspire further innovations in 'prompt engineering' and 'output shaping' for LLMs, focusing on optimizing AI responses for specific user needs and cognitive styles. Community engagement, such as contributions to the skill's rules or forks for different AI agents, will be a key indicator of its long-term impact and adaptability.