70d74e675f
- Rename project name and package: novelWriter -> UmbraEdit, novelwriter -> umbraedit - Rename entry script novelWriter.py -> UmbraEdit.py - Rename package directory novelwriter/ -> umbraedit/ - Rename i18n translation sources nw_*.ts -> ue_*.ts - Rename packaging files (appdata, project xml template) - Update all imports, references, docs, workflows, tests and build config - Update pyproject.toml project name, GUI script and URLs - Preserve full upstream git history; origin repointed below
60 lines
3.2 KiB
Markdown
60 lines
3.2 KiB
Markdown
# AI Policy
|
|
|
|
This is the AI policy of UmbraEdit and associated projects.
|
|
|
|
|
|
### Code Contributions
|
|
|
|
We consider code to be creative work, and we have a strong preference for feature contributions designed by actual
|
|
humans over the average and mediocre output of agentic and generative AI.
|
|
|
|
While we do not have a strict No-AI policy, we want to limit its use in this project. In general, we expect any
|
|
contributed code to hold up to maintainability and stability standards so that our users can rely on the software for
|
|
their creative work. This implies careful review, thorough test coverage, and beta testing of all features before they
|
|
reach live users. The requirements on code that isn't shipped to the user are less strict. This includes unit tests and
|
|
build tooling.
|
|
|
|
We recognise that AI tools can be useful in a developer's toolbox, alongside other tools that can edit or add code. In
|
|
addition to read-only operations, it is generally fine to use them for boilerplate, fixes, and otherwise editing and
|
|
extending existing code. However, you must not contribute fully or predominantly AI-generated features or other
|
|
bulk-generated code. This applies especially to core functionality.
|
|
|
|
In general, so-called vibe coding does not meet our code quality requirements, so this project is not suitable for such
|
|
contributions in any case. The licensing implications around generated code are also unclear, as AI models are trained
|
|
on material without consent or concern for original licensing.
|
|
|
|
No AI automation or agentic tools are used anywhere in this project on GitHub. Do not use your AI agents on your
|
|
contributions on GitHub either. Commits listed in any way as authored by, signed off by, or assisted by, AI tools are
|
|
explicitly not allowed. Pull requests containing such commits will be rejected. You should also not commit AI-specific
|
|
files into the code repository. Most of the common ones are in the git ignore file.
|
|
|
|
|
|
### Design Work
|
|
|
|
No design work generated by AI for UmbraEdit will be accepted. We do not control how third-party work like icon
|
|
libraries are created, but any artwork added to the app must be made by actual people.
|
|
|
|
|
|
### Translation Work
|
|
|
|
This project uses Crowdin to host the translation files for the user interface. Crowdin has a lot of AI integration and
|
|
tooling, which we do not use. The Crowdin translation interface provides AI translated suggestions, which the
|
|
translators may choose to use. All translations are proofread and approved by speakers of the respective languages.
|
|
|
|
|
|
### Application Features
|
|
|
|
We do not believe creative writing and generative AI go together. Therefore, there will not be any tooling or features
|
|
added to this project that will generate text for you. Language models have a real use case in text analysis that
|
|
predates generative AI, and such supportive tools **may** be in scope depending on how intrusive they are.
|
|
|
|
|
|
### Discussions and Comments
|
|
|
|
Please refrain from using AI to generate or respond to discussions, issues, and pull requests. Especially do not use AI
|
|
tools to generate long and verbose issue reports. It is OK to use AI for translations of comments or for accessibility
|
|
reasons. All text in the code or posted to this project must be in English.
|
|
|
|
|
|
*Last updated on 2026-10-03.*
|