What problem does it solve? SparkEngine's build system spans three coexisting configure styles, pinned MSVC toolsets, a machine-checked third-party manifest, and a multi-job GitHub Actions matrix, so configure errors, linker mismatches, and CI failures are hard to diagnose without deep repo knowledge. This Skill provides the verified runbook for building, packaging, and keeping dependencies and CI green. ## Core Features & Use Cases - Build configuration guidance: Choose the correct CMake preset, generator, and build directory layout (presets vs raw -B build vs wrapper scripts) and avoid stale-cache pitfalls with cmake --fresh. - CI failure triage: Map failing jobs (build-windows-vs2022, check-thirdparty-manifest, build-linux-*) to their root causes using the quick triage table and local reproduction commands. - Dependency management: Add or bump third-party dependencies while keeping ThirdParty/dependencies.lock in sync, and handle the SDL2 export repair and header-copy race. - Use Case: A CI run fails with LNK2038 _ITERATOR_DEBUG_LEVEL mismatch. The Skill identifies mixed Debug/Release archives under build/lib, explains the per-config archive output rule, and prescribes nuking build/lib and rebuilding both configs. ## Quick Start Ask the assistant to diagnose why your SparkEngine CMake configure or CI job is failing, mentioning the exact error message or job name.
