levnikolaevich/claude-code-skills

What it solves

It provides a library of specialized engineering "skills" for AI coding assistants (specifically Claude Code and Codex) to help them perform complex software engineering tasks more reliably. Instead of relying on general-purpose prompts, it gives the AI specific procedural guidance, decision gates, and evidence checklists for high-stakes activities like auditing codebases or planning implementations.

How it works

The project is a marketplace of plugins containing standalone skill definitions in Markdown format. These skills are designed to be loaded into the context of the AI model, providing it with domain-specific knowledge and tool guidance. The repository uses a shared layout that allows the same skills to be used across different AI hosts (Claude Code and Codex) via their respective plugin systems.

Who it’s for

Software engineers and maintainers using Claude Code or Codex who want to automate professional-grade workflows such as architectural audits, performance optimization, and test strategy planning.

Highlights

• Review Suite: Tools for validating implementation plans and reviewing completed code from multiple risk perspectives.
• Codebase Audit Suite: Specialized auditors for documentation, general code health, test suites, architecture, and persistence layers.
• Optimization Suite: Tools for profiling performance, upgrading dependencies in verified batches, and modernizing code.
• Testing Suite: Tools for designing risk-based test strategies and building reproducible acceptance tests.
• Product Discovery Suite: Tools for evaluating product opportunities based on evidence.
• Maintainer Suite: A toolkit for managing GitHub projects, including skill reviews and release publishing.