Simplicio Local
A local-first C++ runtime for CPU-only LLM inference with Apple Silicon acceleration, exposed through a unified command-line interface.
Simplicio Local is the local-inference edge of a small "Simplicio" ecosystem of separate repositories: this project provides a vendored CPU engine plus a native C++ runtime and Apple Silicon acceleration path, while sibling repositories cover repository context (simplicio-mapper), a reasoning runtime (simplicio-prompt) and a delivery loop (simplicio-sprint) that feed into it.
It is built with CMake and Ninja and wraps an Ollama or other custom upstream server as a practical chat backend behind a unified CLI offering doctor, plan, convert, chat and serve commands, with documentation covering MLX, Metal, its scheduler, memory and cache behavior. Historical "US4" naming remains in CLI flags, environment variables and file paths for compatibility, though it is not the project's current identity.
Playwright-based CLI smoke tests are described as its main end-to-end test path, and a changelog tracks the CMake project version and the starter package version separately.