Quickstart

This package targets Python 3.12+ and follows a standard src/ layout. See Compatibility And Start Here for the tested package combination and install-path guidance.

Local development setup:

python -m venv .venv
source .venv/bin/activate
make dev
make test

Run the bundled example:

python -m pip install "llama-cpp-python[server]" huggingface-hub
make run-example
make examples-test

make run-example executes the canonical umbrella-level walkthrough that uses a real packaged problem, a live model-backed workflow agent, canonical experiment artifacts, and exported event-table validation.

The walkthrough uses a managed llama.cpp client. Install llama-cpp-python[server] before running it. If you want the default GGUF download path to work, also install huggingface-hub; otherwise set LLAMA_CPP_MODEL to a specific local GGUF file.

make examples-test stays deterministic and offline-first by default. It exercises the two smaller recipe-first examples and skips the live walkthrough unless RUN_LIVE_EXAMPLE=1.

Build the docs:

make docs