C++ Refactoring Assistant Prompt Pack is a Markdown-based collection of 30 structured prompts designed for use with large language models (LLMs) to assist with the analysis, review, and refactoring of existing C++ code. The pack is intended for developers and teams working with legacy, large, or unfamiliar C++ codebases who want a repeatable, structured way to use AI assistance during refactoring work.

The prompts cover C++11 through C++23 and are organized into seven sections. Each prompt is written to produce structured Markdown output, typically including summaries, findings, risk assessments, refactoring plans, patch proposals, test suggestions, and rollback plans.

Contents

• Variables glossary — definitions of all placeholders used across the prompts ([CODE], [TREE], [INCLUDES], [CALL_SITES], [RULES], [ENVIRONMENT], [TESTS],[PLATFORMS], [COMPILERS], and others).

• Master System Prompt — a role and output-format specification that configures the LLM as a conservative senior C++ architect and refactoring assistant.

• Section 1 — Recon & Understanding — codebase recon, file/module analysis, dependency mapping, ownership and lifetime mapping, code smell inventory.

• Section 2 — Architecture & Design — SOLID audit, API redesign, layering review, design pattern evaluation, domain model extraction.

• Section 3 — Modernization — C++ standard migration, raw pointer to smart pointer conversion, macro/template modernization, C-style to RAII/STL conversion, CMake modernization.

• Section 4 — Quality & Safety — readability refactoring, error handling strategy, exception safety audit, undefined behavior and portability audit, thread safety review.

• Section 5 — Performance & Memory — hotspot analysis, allocation and copy reduction, cache locality, benchmarking and profiling plans.

• Section 6 — Testing & Verification — test gap analysis, characterization tests, unit and property test generation.

• Section 7 — Refactoring Execution — step-by-step refactoring plans, automated patch/diff proposals, code review and regression checklists.

Format and Compatibility

• File format: Markdown (.md)

• Language: English

• Compatible with: ChatGPT, Claude, Gemini, and local LLM deployments that accept long-form prompts

• License: Personal and commercial use in the buyer's own projects. Redistribution or resale of the prompt pack itself is not permitted.

Intended Use

The prompts are designed to be used one at a time, with the user supplying the relevant code, project context, C++ standard, build system, constraints, and goal for each task. The Master System Prompt is intended to be placed in a system-prompt field where supported, or prepended to the first user message otherwise. The pack assumes a conservative refactoring workflow in which behavior is preserved unless explicitly changed, changes are made in small verifiable steps, and no refactor is applied without compiling and testing.