Skip to content

bastndev/skills

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

155 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

[Start] / Middle / [End]

Español 🇪🇸 | 中文 🇨🇳 | Deutsch 🇩🇪 | Français 🇫🇷 | 日本語 🇯🇵 | 한국어 🇰🇷 | Português 🇧🇷 | Русский 🇷🇺 | Tiếng Việt 🇻🇳 | हिन्दी 🇮🇳 | العربية 🇸🇦...


Bootstrap new projects with confidence. Iteratively refine and harden them as they grow. Perform deep, safe refactoring when the architecture needs to evolve.

skills.sh

npx skills add bastndev/skills

The Three Phases

Phase Purpose Key Capabilities Example Skills Status
Start Begin a project — scaffold new, or learn from an existing one Create production-ready structures and configs from commit #1. Or analyze an existing codebase's architecture & packages and document it, so you can reuse what works in your own project. start-package, start-astro, skrapi ㅤㅤ✅
Middle Continuous improvement & polish Enhance UI/UX, harden security, boost performance, clean logic, and eliminate dead code during active development. Targeted enhancers (TBD) Planned
End Audit, diagnosis & safe refactoring Full architecture & quality analysis. Categorized findings with file-level evidence. Prioritized phased plan executed only with explicit approval. Behavior-preserving. Multi-runtime support. end ㅤㅤ✅

Available Skills

Listed in the natural order you'd reach for them — Start something new (or study an existing codebase for inspiration), refine it in the Middle, and harden it at the End.

Skill Description
start-package Start — Scaffolds a publishable dual ESM + CJS TypeScript package with bundled type declarations, a zero-config tsup build, and TypeScript pinned to 5.x so it builds cleanly in both CLI and editor. Generates package.json, tsconfig.json, tsup config, a smoke test, and .vscode settings, then installs, builds, and verifies.

Full documentation
start-astro Start — Scaffolds a new Astro project using the minimal template, overlaid with a clean, scalable architecture — ready to grow from a portfolio to a full app. Sets up a shared layout, header, footer, pages, light/dark theme toggle, native View Transitions, path alias, and Content Collections out of the box.

Full documentation
skrapi Start — Point it at a project you admire and it maps how that codebase is built into a fixed SKRAPI/ folder of focused Markdown — architecture, dependencies, and paste-ready prompts — so you can borrow the patterns that fit your own project before you start. Works on any stack (web, mobile, extension, library, monorepo); every description is verified against real code, never guessed from a package name. Multilingual output (🇪🇸 ES · 🇺🇸 EN · 🇨🇳 ZH).

Full documentation
end End — Understands your project end-to-end. Delivers a clear diagnosis (confirmed bugs, risks, opportunities, technical debt) with concrete references, recommends the right architecture direction for this codebase, and builds an ordered execution plan. Every change happens in an isolated, reviewable phase — it only proceeds when you say go, start, or proceed, and no files are touched during analysis.

Full documentation & examples

Note: Each skill ships with its own detailed README. The root page gives the high-level overview; dive into ./skills/<skill>/README.md for deep usage, report examples, and guarantees.

Installation

Every skill installs the same way — pick the one you need:

npx skills add bastndev/skills --skill start-package   # Start  — scaffold a TS npm package
npx skills add bastndev/skills --skill start-astro     # Start  — scaffold an Astro project
npx skills add bastndev/skills --skill skrapi          # Start  — study & document any codebase
npx skills add bastndev/skills --skill end             # End    — audit & safely refactor

How the End Skill Works

end is the most mature skill in the suite. Here's its workflow end to end:

  1. Analysis first — Maps entry points and understands the project. Zero files are modified.
  2. Structured report — Clear findings categorized into Bugs (with severity), Debt/Risks, and Suggestions, plus a scored health overview, an architecture recommendation, and an ordered plan — all backed by concrete file + line references.
  3. You authorize every phase — It executes exactly one phase at a time. After each phase you get a precise summary of changes, validations performed, and the list of remaining phases.
  4. Full control & safety — Never creates tests if the project had none. Never adds dependencies or changes the package manager without permission. Respects your uncommitted work and always preserves current behavior unless fixing a justified bug.

For the complete workflow, exact report formats (including the required closing blocks), architecture decision rules, and all safety guarantees, read the dedicated skill documentation:

End – Refactor Project

The full internal specification lives in skills/end/SKILL.md.

Roadmap

  • Startstart-package (scaffold), start-astro (Astro scaffold) and skrapi (study an existing codebase) ship today; more start-* scaffolders are on the way.
  • Middle — focused, on-demand improvers (performance, security, UX, dead-code removal) are planned.
  • Endend ships today; more runtimes, additional specialized refactoring modes, and utilities are planned.

Each skill ships with its own dedicated documentation (like the current End – Refactor Project).



Built for developers who want their AI agents to act with the discipline of a senior engineer.

If you find any bugs or have feedback, feel free to open an issue.

Made in 🇵🇪 by Gohit X · Licensed under MIT

About

Public repository - Agent Skills (🔥)

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors