<2-3 sentences summarising what the project addresses and solves, the target audience, and the benefit.> Light-skills provides a comprehensive AI-powered "skill package" for Claude Code and Codex, covering the entire research lifecycle from literature review to manuscript submission. It targets students and independent researchers, acting as an AI partner to streamline workflows, ensure reproducibility, and uphold academic integrity by preventing data fabrication.

How It Works

This project integrates 28 specialized AI skills and 9 knowledge bases into AI coding clients, functioning as a cohesive research assistant. Its core innovation is a strict "honesty mechanism" prohibiting fabricated data or literature, marking unverified information. The system provides runnable scripts, templates, and real examples, ensuring actionable and verifiable outputs. "Full-process collaboration" links skills and knowledge bases for consistency, supported by adversarial self-checking and cross-session memory.

Quick Start & Requirements

• Installation: Clone the repository (git clone https://github.com/Light0305/Light-skills.git) and run the install.ps1 (Windows) or install.sh (macOS/Linux) script.
• Prerequisites: Requires Claude Code or Codex, Git. Python/R and LaTeX are recommended.
• Dependencies: Integrates with AI models (Claude Opus 4.8, GPT 5.5 recommended). No specific hardware like GPUs or CUDA is mandated.
• Links: Official GitHub: `https://github.com/Light0305/