Run a structured question and answer session with your coding agent before starting a new feature.
Build a shared vocabulary document so your agent uses the same concise terms as your team.
Adopt a test driven, write a failing test first workflow with your coding agent.
Use a repeatable skill for diagnosing bugs instead of ad hoc debugging prompts.
| loulanyue/skills | zealchou/fable5-scaffold | emadd/mise-en-claude | |
|---|---|---|---|
| Stars | 47 | 47 | 48 |
| Language | Shell | Shell | Shell |
| Setup difficulty | easy | moderate | easy |
| Complexity | 2/5 | 3/5 | 2/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
This repository is a collection of ready made skills for AI coding agents like Claude Code and Codex, meant for real day to day engineering work rather than quick one-off prompting. Each skill is a small, self-contained set of instructions the author uses personally, and the README describes them as easy to adapt and usable with any model. You can install the skills in two ways. The first copies editable files into your own project using a command line installer, so you can open them up and change them to fit your workflow. The second installs them as a managed Claude Code plugin that stays read only and updates automatically whenever the author ships a new version, so you follow along without maintaining your own copy. After installing either way, a setup command asks a few questions, such as which issue tracker you use and where you want generated documents saved. The bulk of the README explains the specific problems each skill is meant to solve. One pair of skills runs a detailed question and answer session with the agent before any code gets written, to close the gap between what you actually want and what the agent assumes you want. Another approach encourages building a short shared vocabulary document for a project, so the agent uses the same concise terms the team already uses instead of long, generic descriptions, which the author says also helps the agent navigate the codebase and think in fewer words. For cases where the agent produces code that does not work, there is a skill that encourages a test driven, write a failing test first then fix it, style of working, plus a separate skill focused on a repeatable process for diagnosing bugs. The README goes on to describe a fourth common problem, code quality degrading into an unstructured mess over time, though the specific fix for that issue was not included in the portion of the README available here.
A collection of reusable AI coding agent skills covering requirement gathering, shared team vocabulary, test driven development, and bug diagnosis.
Mainly Shell. The stack also includes Shell, Claude Code Plugin.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
double-check against the repo, no cap.