A tiny, typed, sequential pipeline engine for Python.

lilpipe is a lightweight, Pydantic-powered library for building and running sequential workflows in Python. It’s ideal for data processing, scientific workflows, and any task that benefits from a clear, linear sequence of steps. With built-in caching and simple control signals, lilpipe is small, typed, and practical.

• Sequential workflows — run steps in a fixed order.
• Type-safe context — Pydantic-based PipelineContext.
• Smart caching — fingerprint_keys to skip unchanged work.
• Flow control — ctx.abort_pass() and ctx.abort_pipeline().
• Composable steps — nest via Step(name, children=[...]).
• Tiny surface area — minimal API and dependencies.

pip install lilpipe

See pyproject.toml for supported Python versions.

Run the full demo notebook live on Binder:

(If you’d rather just view it, the raw notebook is at examples/example.ipynb in the repo.)

Licensed under the Apache 2.0 License.

• Issues: https://github.com/andrewruba/lilpipe/issues
• PyPI: https://pypi.org/project/lilpipe/