AI analysis grounded in the code graph — computed facts, not vibes · 2026-09-23T03:03:03Z
This is a "skill" file (skills/i-have-adhd/SKILL.md) — a set of 10 prompt-level rules — that gets installed into coding agents (Claude Code, Codex, Cursor, Zed, Pi, Hermes) via their plugin/marketplace mechanisms. It instructs the agent to lead with the next action, number steps, cap lists, restate state, and drop preamble/closers, aiming to stop LLMs burying answers under conversational filler. The intended audience is developers using AI coding assistants who want terser, action-first output.
The 1,866-star jump in a single day isn't explained by any tagged release (none fetched) or by codebase complexity — the repo has just 18 symbols, mostly a Python eval script. Growth tracks a burst of commits adding support for more agents (Pi, Hermes, Zed, Cursor via symlink) and marketplace/manifest fixes in the last 30 days, which broadens where the skill can be installed, but the scale of the spike looks driven by the relatable framing and shareability of the README's before/after example rather than by any measurable engineering milestone.
What changed recently, how it's actually built (from the code graph), and whether you should care. Free account — no card, no spam.