levnikolaevich/claude-code-skills

What it solves

This project provides a library of standalone engineering "skills" designed to enhance the capabilities of AI coding agents like Claude Code and Codex. It solves the problem of providing AI agents with structured, domain-specific knowledge and rigorous checklists for complex software engineering tasks—such as architecture auditing, performance optimization, and test strategy planning—without requiring the agent to guess the best approach.

How it works

The project implements the Agent Plugins v1 standard, allowing these skills to be portable across different AI hosts. Each skill is a compact, bounded set of instructions (under 200 lines) that defines a specific outcome, provides domain knowledge, and requires the agent to collect evidence and perform self-checks before reaching a verdict. These skills are organized into suites (e.g., Review, Audit, Optimization, Testing, Architecture) and can be installed as plugins into Claude Code or Codex.

Who it’s for

Software engineers and architects who use AI coding agents and want to ensure their agents follow professional engineering standards, rigorous audit processes, and structured design patterns when modifying or auditing a codebase.

Highlights

• Portable Standard: Supports the Agent Plugins v1 specification for cross-platform compatibility.
• Comprehensive Suite: Includes specialized skills for codebase auditing, system design baseline building, performance optimization, and delivery review.
• Detailed Evidence-Based Approach: Each skill requires the agent to track checklist items with evidence or explicit reasons for non-applicability.
• Modular Installation: Users can install only the specific suites they need (e.g., just the architecture-suite or review-suite).