Third-party notices¶
ParleML is built on open-source software and includes some open data. Each of those components stays under its own licence, and its notice travels with every copy of ParleML. ParleML's restrictions do not override rights granted by those licences. When redistributing scoring packages, include the ParleML license and the notices required by the components you distribute.
| Where you got ParleML | Where the notices are |
|---|---|
| The container image | /opt/parleml-licenses/ inside the image. Read them with docker run --rm ghcr.io/parleml/parleml:0.8.0 ls /opt/parleml-licenses. |
The Python packages (parleml-ml4, parleml-mli) |
LICENSE.txt and NOTICE.txt in the package's metadata folder (pip show -f parleml-ml4 lists them). NOTICE.txt covers what the package itself contains; every dependency pip installs carries its own licence. |
| This website | The files below. |
The files¶
- ParleML terms of use: the terms as plain text.
- Python components: every Python package in the image, with its full licence text.
- Web app components: every package in the browser app, with its full licence text.
- Notices for users: what to know about IP geolocation (you bring your own MaxMind database), postal-code data, the optional name corpus and AI features.
- Hand-kept notices: vendored code (RuleCOSI, MIT), bundled data (GeoNames postal codes, CC BY 4.0), and the licence texts of components that ship without one (Codex CLI, Treelite, the Apache-2.0 text).
The image's agent tools, Codex CLI and Gemini CLI (Apache-2.0), and Node.js (MIT) keep their licences in the image beside the notices. Claude Code and Cursor are not included, because their licences don't allow it; the extended image installs them on your machine, under their own terms.