51K Developers Fixed AI Verbosity Before Claude Did

A coding assistant gives you three paragraphs about why your question is interesting, two more about approaches it considered but rejected, then finally—buried in paragraph six—the four lines of code you actually needed.

This problem annoyed developers enough that 51,000 of them starred a Markdown file designed to stop AI coding assistants from burying answers in apologetic preambles. Months later, Anthropic shipped a native "Concise" mode in Claude Code that does essentially the same thing. Two solutions, same pain point, zero rivalry.

The Problem: AI That Explains Before It Helps

The i-have-adhd repository targets a specific frustration: responses that bury actionable information beneath preambles, tangents, repeated state, vague estimates, and closing pleasantries. Hacker News threads echo the complaint—Claude producing verbose explanations of rejected approaches, irrelevant negative statements, and long responses where the useful result is difficult to find.

You ask for a regex pattern. You get a history lesson on regular expressions, three cautionary notes about edge cases, then the pattern. For developers with ADHD or anyone juggling context-heavy workflows, this verbosity isn't just annoying—it costs mental energy.

The Community Solution: A Markdown Skill That Works Everywhere

The i-have-adhd project is an instruction set—a skill file you drop into your coding agent's configuration. It works across nine different platforms: Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Kimi Code, OpenCode, Qwen Code, and Pi. No vendor lock-in, no proprietary API.

The momentum is real. Beyond 51,400 stars and about 3,000 forks, the repository expanded into integrations, hooks, evaluations, benchmarks, translations, and plugins for multiple coding-agent platforms. An independent Reddit project called bury-bench even uses the repository's publicly documented ADHD-friendly guidelines as the basis for deterministic evaluation of coding-agent responses.

When a workaround spawns its own benchmarking framework, the problem is real.

The Product Solution: Anthropic Builds It Native

Claude Code's Concise output style places results first, removes preamble and narration, and keeps responses short while preserving full detail when requested. Same philosophy, baked into the product. No configuration file to maintain, no cross-platform compatibility issues.

This isn't a case of a big company stealing an open-source idea. Anthropic didn't need to see the i-have-adhd repo to know their users wanted less fluff. The independent arrival at the same solution validates that the problem mattered enough for professional product development.

Convergent Evolution, Not Competition

Both approaches deserve respect. The open-source project proved market demand and delivered cross-platform compatibility. The native implementation offers straightforward integration for Claude users. Neither cancels out the other.

The i-have-adhd repository has open issues around performance concerns involving unnecessary I/O, context overhead, and evaluation runtime—typical growing pains for a project scaling this fast. Twenty-eight issues and forty-eight pull requests show the maintainers are actively refining it.

What This Means for Developer Tools

When 51,000 developers rally around a Markdown file that just tells AI to get to the point, they're signaling that default verbosity is a broken assumption. Some will prefer the flexibility of a cross-platform skill file. Others will choose the polish of a native feature. Both paths are legitimate.

The real lesson: user workarounds this popular are product requirements in disguise. The verbosity problem is getting the attention it deserves—whether you're running a community solution or using Claude Code's built-in mode, you're no longer stuck reading AI apologies before you get your code.