TheQtCompanyRnD/agent-skills

What it solves

It provides a standardized set of "agentic skills" to help AI coding assistants (like Claude Code, GitHub Copilot, and Gemini CLI) better understand and work with the Qt framework. It specifically addresses common LLM failures, such as pre-training biases toward legacy tools (qmake) or incorrect assumptions about QML bindings and scoping.

How it works

Skills are packaged as directories containing a SKILL.md file with YAML frontmatter and optional reference documentation. This structure allows AI agents to use "progressive disclosure," loading only basic metadata at startup and pulling in detailed technical references only when the specific skill is activated. The project also includes an MCP (Model Context Protocol) server for real-time Qt API documentation lookup.

Who it’s for

Software developers and QA engineers using AI coding tools to develop applications with Qt, QML, and C++.

Highlights

• Code Review: Specialized skills for C++ and QML that combine deterministic linting with deep AI analysis for memory safety, thread safety, and performance.
• Conceptual Corrections: Guides that correct LLM biases regarding declarative QML and the C++/QML boundary.
• Automation Tools: Skills for generating QML test cases, extracting Figma design tokens into QML singletons, and managing CMake projects.
• Cross-Platform Support: Compatible with Claude Code, Codex CLI, Gemini CLI, and GitHub Copilot.
• Documentation Integration: A hosted MCP server for direct API documentation access.