all-skills

AI Agent Skills Collection

A curated set of AI Agent Skills to boost productivity and code quality for AI coding assistants (Cursor, Claude Code, etc.).

Included Skills

Project Setup

Quickly configure development toolchains, code standards, and AI agent skills for new projects. Automatically detects the project environment and adapts to the host project's package manager, framework, and existing toolchain.

• Smart environment detection (package manager, project type, TypeScript)
• Auto-generate AGENTS.md project conventions document
• Configure Prettier code formatting and Vitest test framework
• Install popular AI Agent Skills (Vercel React, Frontend Design, shadcn/ui, Vitest)
• Configure next-intl internationalization for Next.js projects

Documentation →

npx skills add https://github.com/Yukiniro/skills --skill project-setup

Prompt Optimizer

Transform vague or simple user prompts into high-quality, structured AI instructions. Uses systematic optimization techniques like XML tagging, few-shot examples, and Chain-of-Thought.

• Systematic prompt optimization based on the "AI as a New Employee" philosophy
• Structured prompt construction (role, context, task, requirements, output_format)
• Improve reliability, accuracy, and format consistency of AI outputs

Documentation →

npx skills add https://github.com/Yukiniro/skills --skill prompt-optimizer

Frontend Resume

Convert PDF resumes into zero-dependency, animation-rich interactive single-page HTML resume websites. Help users discover their preferred aesthetic through visual exploration.

• Zero-dependency single-file output (inline CSS/JS, no build tools)
• 10 distinctive style presets (Midnight Architect, Neon Terminal, Clean Slate, etc.)
• Scroll-triggered animations, floating navigation, interactive skill bars
• Responsive layout, print-friendly, WCAG AA accessibility support

Documentation →

npx skills add https://github.com/Yukiniro/skills --skill frontend-resume

Resume Screener

Systematically evaluate resumes against job requirements using a 10-level grading system (C → SSS) with 9-dimension multi-angle scoring, producing evidence-driven evaluation reports and hiring recommendations.

• 9-dimension evaluation (education, work experience, technical skills, project experience, etc.)
• Supports Chinese and English resumes, understands 985/211, BAT/FAANG systems
• Hard requirement detection with grade capping mechanism
• Structured output: overall rating, dimension details, strengths/weaknesses analysis, interview suggestions

Documentation →

npx skills add https://github.com/Yukiniro/skills --skill resume-screener

Work Weekly Report

Generate structured weekly work summaries from daily work logs. Groups work by project/topic instead of by day, extracting objective facts with categorized dimensions.

• Group by project/topic, not chronological order
• Objective facts only — no subjective evaluations
• Categorize by dimension (Business/Feature, Technical/Infrastructure, Data/Research)
• Supports bilingual output (Chinese/English)

Documentation →

npx skills add https://github.com/Yukiniro/skills --skill work-weekly-report

Smart Commit

Automatically analyze git changes and create Conventional Commits with Chinese descriptions. Fully automatic — analyzes diff, determines type, generates message, and commits without confirmation.

• Conventional Commits format: type(scope): 中文描述
• Auto type detection (feat, fix, docs, refactor, perf, test, build, ci, style, chore, revert)
• Smart staging: uses existing staged changes or stages all if empty
• Concise Chinese commit descriptions (under 50 characters)
• Sensitive file detection (.env, credentials) — warns and blocks commit
• Respects pre-commit hooks (no --no-verify)

Documentation →

npx skills add https://github.com/Yukiniro/skills --skill smart-commit

Squash Commits

Squash all commits on the current feature branch (relative to main/master) into a single clean commit while preserving the branch name. Includes safety checks, user confirmation, and smart commit message generation.

• Auto-detect main branch (main/master)
• Safety checks: clean working directory, not on main, >=2 commits required
• Smart commit message with all original messages as bullet points
• Post-squash guidance (force-push reminder)
• Bilingual triggers (English + Chinese)

Documentation →

npx skills add https://github.com/Yukiniro/skills --skill squash-commits

Deep Code Analysis

Systematically analyze and understand code logic and business requirements. Follows a "measure first, understand second, act last" principle with a 6-step workflow.