Sign InOpen Brain
GitHubGitHub RepoNeeds Review

ayghri/i-have-adhd

i-have-adhd is a coding-agent skill that enforces action-first, compact responses with numbered steps and a concrete next action. It is a useful prompt-policy example for reducing agent verbosity.

GitHub · Trending today
Open Source Open MarkdownOpen JSON
Source Summary

The repository provides an **ADHD-friendly coding-agent skill** built around **10 response rules**. It removes preambles and closers, puts the action first, numbers steps, uses specific time estimates, and ends with one concrete next step.

Practical Implication

Builders can study or adapt the skill as a compact output contract for coding agents, especially when default responses bury commands or file edits in explanation. Its before-and-after example favors executable guidance and a clear failure handoff.

Agent-Ready Context
The repository provides an **ADHD-friendly coding-agent skill** built around **10 response rules**. It removes preambles and closers, puts the action first, numbers steps, uses specific time estimates, and ends with one concrete next step.

Builders can study or adapt the skill as a compact output contract for coding agents, especially when default responses bury commands or file edits in explanation. Its before-and-after example favors executable guidance and a clear failure handoff.

The approach optimizes presentation rather than model correctness. It is loosely informed by an ADHD toolkit but is explicitly an adaptation for LLM responses, not a diagnosis or general system for organizing human work.
Connected Context · Feed7 Judgment

This narrows agent-skill design to a compact output contract: action ordering, numbered execution, concrete timing, and one next step. Unlike broader engineering workflow packs, it improves how guidance is presented rather than how work is verified or whether it is correct, so its value should be judged through usability and task completion rather than model-quality claims.

addyosmani/agent-skillsThe ADHD-friendly skill governs response presentation, while this pack governs engineering lifecycle gates and evidence; together they separate readable execution guidance from verification discipline.Direct agents with visual prompts in Design ModeDesign Mode improves how users supply visual intent to an agent, whereas this skill improves how the agent returns actionable guidance, addressing opposite sides of the developer interaction.SkillProx: Self-Evolving Agent Skills via Proximal Textual Gradient DescentSkillProx supplies a measured edit, retry, rollback, and pruning process that could evaluate and maintain these ten rules instead of assuming their continued usefulness.anthropics/claude-codeClaude Code’s plugin and custom-command support provides a plausible host for this portable response contract, but the skill itself does not add Claude Code’s workflow or correctness controls.
Context Map
agentcoding#skills#prompting#dev-ux
Uncertainty
The approach optimizes presentation rather than model correctness. It is loosely informed by an ADHD toolkit but is explicitly an adaptation for LLM responses, not a diagnosis or general system for organizing human work.