giotto-tda, the scikit-learn-compatible TDA toolkit, and the corpus's deliberate out-of-profile L0 fixture.

giotto-tda

giotto-tda offers persistence, Mapper, and time-series topology behind a scikit-learn transformer API, which makes it convenient inside an ordinary ML pipeline.

It is the corpus’s intentional L0 fixture. giotto-tda is wheel-only on PyPI — no sdist and no conda package — so the manifest reaches it through [pypi-dependencies] and falls out of the biopixi profile entirely. This is a packaging fact, not a judgement about the library. Adding a recipe under recipes/giotto-tda and switching to a path dependency would promote it to L1; nothing about the software prevents that.