# Are We AI Friendly?

> Where open source projects stand on AI: code contributions, issues and AI-written summaries, rated from each project’s own policy.

Each project has an overall stance derived from its rules for AI-assisted code contributions, plus a stance on the same scale for AI-found issues and AI-written text. Every stance other than "No AI policy" is backed by a verbatim quote from the project's own policy.

Stances:
- **AI-friendly** (shown as Friendly): AI-assisted contributions are welcome under the same rules as any other contribution.
- **AI-friendly with conditions** (shown as Mostly friendly): AI-assisted contributions are accepted with extra rules, such as disclosing AI use, or AI is banned for issues or written text.
- **Not AI-friendly** (shown as Unfriendly): AI-generated contributions are not accepted; at most minor help such as autocomplete or grammar fixes.
- **No AI policy** (shown as No policy): No published policy on AI-assisted contributions.

## Projects

- [redot-engine](https://areweaifriendly.com/projects/redot-engine.md): AI-friendly with conditions. Redot Engine explicitly allows contributors to use LLMs and agentic coding assistants, as long as the human understands and takes responsibility for the work, tests pull requests manually before review, discloses AI involvement in art assets, and keeps licensing clear. Purely AI-generated artwork is not permitted, and unreviewed, low-effort or spammy AI output may be rejected.
- [Alacritty](https://areweaifriendly.com/projects/alacritty.md): Not AI-friendly. Contributions must not include content generated by LLMs or other probabilistic tools. This covers code, documentation, pull requests, issues and comments.
- [Apache Airflow](https://areweaifriendly.com/projects/apache-airflow.md): AI-friendly with conditions. Gen-AI tools may be used to create pull requests if contributors follow the ASF guidance, review and test all generated code, state AI use in the PR description and take final responsibility. Blindly pasted AI code leads to closed PRs and possible blocking.
- [Apache Arrow](https://areweaifriendly.com/projects/apache-arrow.md): AI-friendly with conditions. AI tools may be used to generate PRs, but authors should review and own every change, be upfront about what was AI-generated, and disclose any copyrighted material. PRs that look fully AI-generated with little author engagement may be closed, and AI agents should never ping maintainers.
- [Apache CouchDB](https://areweaifriendly.com/projects/apache-couchdb.md): Not AI-friendly. Contributing material generated by AI or LLMs is expressly forbidden, including code, documentation and commit messages, because such tools cannot assure Apache 2 licence compatibility or proper attribution.
- [Apache DataFusion](https://areweaifriendly.com/projects/apache-datafusion.md): AI-friendly with conditions. AI-assisted PRs are welcome from anyone, but not unreviewed "AI dumps". Authors must read the whole PR, understand the core ideas, flag parts they do not fully understand and follow the ASF licensing guidance. Replies to reviewers must be in the contributor's own words, though AI may help understand or translate.
- [Apache Software Foundation](https://areweaifriendly.com/projects/apache-software-foundation.md): AI-friendly. Code or documentation generated wholly or partly with AI can be contributed if the tool's terms allow open source licensing and any third-party material in the output is permitted and compatible. Naming the tool in the commit message, for example with Generated-by, is recommended. Individual projects may add their own rules.
- [Asahi Linux](https://areweaifriendly.com/projects/asahi-linux.md): Not AI-friendly. Generative AI tooling is broadly forbidden for material contributions, citing ethical concerns and the clean-room rules of reverse engineering. An LLM-drafted GitHub issue may simply be closed, and concealing extensive LLM use leads to an immediate ban.
- [Astropy](https://areweaifriendly.com/projects/astropy.md): AI-friendly with conditions. Contributions made with generative AI are welcome. Contributors are responsible for and must be able to explain all changes, must describe substantive AI use, and must answer reviewers themselves rather than relaying AI replies. Autonomous agents are not accepted.
- [Bevy](https://areweaifriendly.com/projects/bevy.md): AI-friendly with conditions. Careful, human-driven AI use is tolerated for code if it is disclosed and the contributor understands every line. AI may not write documentation, issues, PR descriptions or release notes, AI-generated media is banned, and AI output may not be pasted into conversations except to help with translation or accessibility.
- [Cataclysm: Dark Days Ahead](https://areweaifriendly.com/projects/cataclysm-dda.md): Not AI-friendly. All content sourced from LLMs is not allowed, including code, configuration, issue and PR text, research and testing results, because it has no authorship and questionable provenance.
- [chezmoi](https://areweaifriendly.com/projects/chezmoi.md): Not AI-friendly. Using an LLM to make any kind of contribution leads to an immediate ban without recourse.
- [Cilium](https://areweaifriendly.com/projects/cilium.md): AI-friendly with conditions. Generative AI may help explain code, autocomplete, refactor, brainstorm and draft tests, as long as contributors review everything, take full responsibility, respect licensing and report non-trivial AI use. Messages substantially written by AI are not acceptable in community spaces, though translation help is understood.
- [Clojure](https://areweaifriendly.com/projects/clojure.md): Not AI-friendly. Code generated by LLMs is not acceptable because it does not comply with the Contributor Agreement. Clojure's code is written and reviewed by humans.
- [CloudNativePG](https://areweaifriendly.com/projects/cloudnativepg.md): AI-friendly with conditions. AI-generated code, docs or comments are accepted if the contributor fully understands and can explain every line, guarantees licensing and provenance, and discloses significant AI use with an Assisted-by trailer. Unvetted boilerplate, AI-hallucinated fixes, unsolicited AI refactors and vague AI-written PR descriptions are prohibited.
- [conda](https://areweaifriendly.com/projects/conda.md): AI-friendly with conditions. Generative AI tools are welcome. Contributors are responsible for reviewing and understanding AI output, must respect the licence and CLA, and may not use agents to submit or review autonomously. In review, contributors should engage directly rather than paste AI responses.
- [Debian](https://areweaifriendly.com/projects/debian.md): AI-friendly. By general resolution in August 2026, Debian chose "Responsible Use of Generative AI" over seven other options, including bans. Generative AI is neither endorsed nor prohibited: contributors must understand, review and test AI-assisted output, consider licensing and remain fully accountable. Disclosure is encouraged but not required, and mass automated actions need prior discussion and human oversight.
- [Django](https://areweaifriendly.com/projects/django.md): AI-friendly with conditions. AI tools may be used if contributors disclose which tools they used and for what, and review and verify all output. Unverified AI-generated contributions may be closed without review, and automated AI reviews must not be requested on Django pull requests.
- [Drupal](https://areweaifriendly.com/projects/drupal.md): AI-friendly with conditions. AI tools may be used if contributors fully verify the code, can explain every decision, ensure GPL compatibility and disclose significant AI-generated code or text. AI output should not be pasted verbatim into issues and comments; contributors should use their own words.
- [EasyBuild](https://areweaifriendly.com/projects/easybuild.md): AI-friendly with conditions. Contributors must declare any AI use, naming the tool, model and extent, and keep a human in the loop. This covers pull requests, issues, documentation, chat messages, mailing list posts and talks.
- [Endless Sky](https://areweaifriendly.com/projects/endless-sky.md): Not AI-friendly. AI-generated or AI-assisted content is forbidden in Endless Sky development, for ethical, philosophical and legal (GPLv3) reasons.
- [FastAPI](https://areweaifriendly.com/projects/fastapi.md): AI-friendly with conditions. Contributors are encouraged to use any tools, including AI, as long as there is meaningful human judgement and effort; PRs that cost more to review than to prompt should not be sent. PRs that look AI-generated or automated are closed, and LLM-generated comments and descriptions should not be pasted.
- [Firefox](https://areweaifriendly.com/projects/firefox.md): AI-friendly. Contributors may use AI tools when writing Firefox code, but AI-assisted patches must meet the usual standards. Contributors must understand, explain and fully self-review every change and remain accountable for it.
- [Flutter](https://areweaifriendly.com/projects/flutter.md): AI-friendly. AI-assisted PRs are accepted if the contributor reviews all generated code, understands and can discuss it, and submits it as their original work under the CLA. AI-generated issue text and PR descriptions must be verified and edited, and unsupervised agents may not post in the issue tracker.
- [Forgejo](https://areweaifriendly.com/projects/forgejo.md): Not AI-friendly. Code, documentation and other works partly or fully generated by AI are not accepted, and any AI help must be disclosed. Commit messages, PR text, comments and issues must not be AI-generated apart from translation and spell-checking, and AI review, vibe coding and agent mode are not allowed.
- [GCC](https://areweaifriendly.com/projects/gcc.md): Not AI-friendly. Legally significant contributions that include or derive from LLM-generated content are declined, except for test cases; small LLM-generated changes may be accepted if clearly marked with Assisted-by. AI may be used for research, bug finding, review and debugging if its output stays out of the contribution.
- [GDAL](https://areweaifriendly.com/projects/gdal.md): Not AI-friendly. LLMs may only be used as improved autocompletion or for mechanical refactoring; humans must be the primary authors. Vibe-coded contributions and agent submissions are banned, and any LLM use in code, tickets or commit messages must be labelled.
- [gedit](https://areweaifriendly.com/projects/gedit.md): Not AI-friendly. Contributions generated by LLM tools and chatbots are not allowed anywhere in the project. The only exception is translating issue and comment text into English.
- [GIMP](https://areweaifriendly.com/projects/gimp.md): Not AI-friendly. Contributors without a developer role must declare in every merge request that it contains no AI-generated content: no code, text or images, only human-created work.
- [Gitea](https://areweaifriendly.com/projects/gitea.md): AI-friendly with conditions. AI-assisted contributions are welcome if AI use is clearly disclosed and contributors review and test the code and can explain and defend it themselves. AI may help draft issues and PRs, but questions from maintainers must be answered without AI.
- [GNU Binutils](https://areweaifriendly.com/projects/gnu-binutils.md): Not AI-friendly. LLM-generated patches are not accepted because their copyright status is unclear. LLMs may help without generating code, and legally insignificant changes such as spelling fixes are fine. Every submission must involve a human; fully automated submissions are refused.
- [Godot Engine](https://areweaifriendly.com/projects/godot.md): Not AI-friendly. The Godot Foundation announced that all code must be human-authored: AI may help only with menial things like code completion, regex or find and replace, and any AI use must be disclosed. Autonomous agents and vibe coding lead to a ban, and AI-generated text is not allowed in issues, PRs or proposals, though machine translation of human writing is fine.
- [GoToSocial](https://areweaifriendly.com/projects/gotosocial.md): Not AI-friendly. Changes created with the aid of AI tooling, code or otherwise, are not accepted, on ethical grounds.
- [gVisor](https://areweaifriendly.com/projects/gvisor.md): AI-friendly with conditions. Contributors are expected to embrace AI tools, with mandatory human oversight: all AI-generated content must be reviewed, the contributor remains the sole responsible author, and substantial AI content must be labelled, for example with Assisted-by. This applies to code, issues, security reports and comments.
- [Home Assistant](https://areweaifriendly.com/projects/home-assistant.md): AI-friendly with conditions. Under the Open Home Foundation policy, AI may be used for code and docs if the contributor reviews, understands and can explain every change. Autonomous agents are not allowed. AI may help with writing, but answers to maintainers must be the contributor's own, and quoted AI output must be marked.
- [Homebrew](https://areweaifriendly.com/projects/homebrew.md): AI-friendly with conditions. Issues and pull requests made with AI are allowed if the tool or model is disclosed up front and all generated content is reviewed first. AI may not be credited in commits, contributors must answer maintainers themselves without AI, and non-maintainers may have only one AI-assisted PR open at a time.
- [Jellyfin](https://areweaifriendly.com/projects/jellyfin.md): AI-friendly with conditions. LLMs may assist with code, but pure vibe coding is rejected: contributors must review, test and explain their changes and handle review themselves. LLM output is prohibited in issues, feature requests, PR bodies, comments and forum posts, except clearly noted LLM translations.
- [Joomla!](https://areweaifriendly.com/projects/joomla.md): AI-friendly with conditions. AI may be used, but contributors must fully understand, review and test all AI-generated code, ensure GPL compatibility, and label AI-assisted PRs with [AI]. Vibe-coded submissions are not allowed, and AI-generated explanations and comments must be reviewed; raw AI answers are discouraged.
- [Krita](https://areweaifriendly.com/projects/krita.md): Not AI-friendly. Krita has a moratorium on AI tools in development because its community has not reached consensus; until a decision is made, using AI when working on Krita is not allowed. The moratorium was set to run until October 2026.
- [Kubernetes](https://areweaifriendly.com/projects/kubernetes.md): AI-friendly with conditions. AI tools may help write a PR if the author understands every change, verifies it first and discloses AI use in the PR description. Large AI-generated PRs, AI-generated commit messages and AI co-author trailers are not allowed, and review comments must be answered without AI.
- [libadwaita](https://areweaifriendly.com/projects/libadwaita.md): Not AI-friendly. Contributions generated by LLMs and chatbots are banned in all parts of the project, including code, documentation, issues and artwork. Translating issue and comment text into English is the only exception.
- [libxml2](https://areweaifriendly.com/projects/libxml2.md): Not AI-friendly. A strict no-LLM policy: no LLMs for issues, patches or pull requests, or for bug tracker comments, including translation.
- [Linux Foundation](https://areweaifriendly.com/projects/linux-foundation.md): AI-friendly. Code or other content generated wholly or partly with AI tools can be contributed to Linux Foundation projects. Contributors must check that the tool's terms do not conflict with the project's licence and that any third-party material in the output is properly licensed and attributed. Individual projects may set stricter rules.
- [Linux man-pages](https://areweaifriendly.com/projects/linux-man-pages.md): Not AI-friendly. Any content created or derived with the help of AI tools is forbidden, including output of AI linters, static analyzers and summarizers. Only accessibility tools such as screen readers and text-to-speech are allowed.
- [LLVM](https://areweaifriendly.com/projects/llvm.md): AI-friendly with conditions. Contributors may use any tools, but a human must read and review all LLM-generated code or text, stay fully accountable, and label substantial tool-generated content, for example with Assisted-by. Agents acting without human approval are banned, and AI tools may not be used on good first issues.
- [LÖVE](https://areweaifriendly.com/projects/love.md): Not AI-friendly. Pull requests, bug reports and other contributions made with LLM or generative AI technology are not accepted.
- [Matplotlib](https://areweaifriendly.com/projects/matplotlib.md): AI-friendly with conditions. AI may help contributors understand code, get ideas and proof-read text, but they must fully understand the output and not solve problems they could not solve themselves. AI use is described in the PR template, PR summaries must be in the contributor's own words, and bots or agents may not interact with the project.
- [Mesa](https://areweaifriendly.com/projects/mesa.md): AI-friendly with conditions. AI-assisted code is allowed if the submitter understands it, takes responsibility including MIT licence compatibility, and discloses AI involvement with Assisted-by or Generated-by. Autonomous tools may not submit or interact with issues or MRs, and AI should not write comments, commit messages or GitLab comments.
- [mGBA](https://areweaifriendly.com/projects/mgba.md): Not AI-friendly. AI-generated code is not accepted; pull requests with AI-generated code are summarily closed.
- [MicroPython](https://areweaifriendly.com/projects/micropython.md): AI-friendly with conditions. Generative AI coding tools are welcome if contributors read and validate the code and talk to reviewers themselves. Copy-pasting to and from a chatbot during review is unacceptable except for translation, and AI agents may not submit pull requests on their own.
- [napari](https://areweaifriendly.com/projects/napari.md): AI-friendly with conditions. Any tools, including AI, may be used, but contributors are responsible for their changes and must understand, read and test them. AI agents may not submit PRs. LLM-written messages are acceptable if the author has read and stands behind them, with quoted LLM answers marked as quotes.
- [Neovim](https://areweaifriendly.com/projects/neovim.md): AI-friendly. Using AI for contributions is acceptable as long as the contributor reviews the output before requesting review and trims verbosity from code, docs, comments, PR descriptions and commit messages.
- [NLnet Labs (Unbound, NSD)](https://areweaifriendly.com/projects/nlnet-labs.md): Not AI-friendly. All code and documentation must be human-authored with no LLM output. LLM-found vulnerabilities and bug reports are accepted if a human verifies them and discloses LLM use, and LLMs may be used for linting, analysis and review. PR descriptions must be in the contributor's own words.
- [Node.js](https://areweaifriendly.com/projects/nodejs.md): AI-friendly with conditions. AI tools are not prohibited, but contributors should disclose AI-generated contributions and how they verified them, take full responsibility and meet the DCO and licensing rules. Automated PRs need prior approval, AI may not fix good first issues, review responses must not be automated, and messages generated entirely by AI should not be pasted.
- [NumPy](https://areweaifriendly.com/projects/numpy.md): AI-friendly with conditions. AI-assisted pull requests are accepted if AI use is disclosed in detail and the author understands and can explain the code. Autonomous agents are not allowed, and AI may not write descriptions, comments or issues except for translation or grammar fixes.
- [Nura (postmarketOS)](https://areweaifriendly.com/projects/nura.md): Not AI-friendly. Generative AI tools are forbidden: contributions fully or partly created with them are not allowed, nor is recommending them to other community members. The project cites ethical concerns and the extra burden such contributions put on maintainers.
- [OCaml](https://areweaifriendly.com/projects/ocaml.md): AI-friendly with conditions. AI-assisted contributions are judged on merit. Contributors must have written or reviewed every part, take responsibility for provenance, and disclose significant AI-generated code, PR descriptions or comments with the tool used. Bug reports should contain only human-verified information, with any LLM content clearly separated.
- [OpenJDK](https://areweaifriendly.com/projects/openjdk.md): Not AI-friendly. An interim policy bans content generated in part or in full by LLMs or similar systems, covering code, text, images, pull requests, e-mail, wiki pages and issues. Contributors may still use AI privately to understand, debug and review code, as long as none of its output is contributed.
- [OpenStreetMap website](https://areweaifriendly.com/projects/openstreetmap.md): AI-friendly. AI tools may be used to help create pull requests if a human has reviewed, tested and fully understands all the changes. Issues tagged for new contributors should not be solved with AI.
- [OpenTTD](https://areweaifriendly.com/projects/openttd.md): Not AI-friendly. Issues and pull requests generated by LLMs are not accepted, and using AI to generate entire lines of code is unacceptable. AI may proofread issue and comment text but not write it.
- [Pallets (Flask, Jinja, Click)](https://areweaifriendly.com/projects/pallets.md): Not AI-friendly. New contributors may not use LLM or AI tools at all, and contributions that look AI-generated are closed with the author likely blocked. Trusted contributors' work must still be understood and created personally. The policy covers all interactions, and AI may not be used to translate or improve what you write.
- [pandas](https://areweaifriendly.com/projects/pandas.md): AI-friendly with conditions. AI coding assistants may help write code, tests and docs if their use is disclosed with the specific tool and the output is fully reviewed. In issues, pull requests and review comments AI may not speak for contributors, apart from translation and grammar editing.
- [pip](https://areweaifriendly.com/projects/pip.md): AI-friendly with conditions. LLM tools may be used, but a human must own the copyright, understand the code and take full responsibility for every line. LLMs may not be listed as co-authors and unsupervised agents are banned. LLM-generated comments must be accurate and stood behind.
- [Polars](https://areweaifriendly.com/projects/polars.md): AI-friendly with conditions. All AI use must be disclosed with the tools used, AI-generated code must be fully tested by a human and is only accepted for accepted issues, and AI may not be used on good first issues. Agents may not interact with the repository at all, and AI-assisted issues and discussions must be reviewed and edited by a human.
- [Processing](https://areweaifriendly.com/projects/processing.md): Not AI-friendly. Fully AI-generated pull requests are not accepted. AI may be used only assistively, like a reference or tutor, not to write code; contributors must understand, test and take responsibility for every line and note which AI tools they used.
- [pytest](https://areweaifriendly.com/projects/pytest.md): AI-friendly with conditions. Contributions made with AI tools are welcome as long as the submitter understands the changes, can respond to review and takes responsibility. Purely agentic PRs without meaningful human review are closed and their authors banned. Crediting AI with Co-authored-by is appreciated but optional.
- [PyTorch](https://areweaifriendly.com/projects/pytorch.md): AI-friendly with conditions. AI tools are supported for issues, pull requests, reviews and comments. A human must understand and take responsibility for the work, and AI-generated text must be clearly labelled and paired with human commentary. Fully autonomous agents are not accepted.
- [QGIS](https://areweaifriendly.com/projects/qgis.md): Not AI-friendly. No LLM-, AI- or agent-assisted code contributions are allowed in the main QGIS repository. Contributions later found to have used AI face code reversion and sanctions on contributor rights.
- [Redox OS](https://areweaifriendly.com/projects/redox-os.md): Not AI-friendly. Contributions generated by LLMs are not accepted.
- [Requests](https://areweaifriendly.com/projects/requests.md): AI-friendly with conditions. LLM tools may be used, but a human must hold the copyright, understand the code and take full technical and legal responsibility. LLMs may not be co-authors, unsupervised agents are banned, LLM-generated review comments are not wanted, and summaries must be fully fact-checked.
- [Rocky Linux Documentation](https://areweaifriendly.com/projects/rocky-linux.md): AI-friendly with conditions. AI assistance is allowed in the documentation project, based on Fedora's policy. Contributors are accountable for quality and licence compliance and must disclose AI use when a significant part comes from a tool unchanged, for example with Assisted-by or document metadata.
- [Ruff](https://areweaifriendly.com/projects/ruff.md): AI-friendly with conditions. AI is supported as a coding tool, with contributors responsible for what they submit and a human who understands the work in the loop. Autonomous agents are not allowed. Issues, PR bodies and replies must be in the contributor's own words; quoted AI context must be marked and explained, and AI editing for non-native speakers is accepted.
- [Rust](https://areweaifriendly.com/projects/rust.md): Not AI-friendly. In rust-lang/rust, cargo, clippy and other adopting repositories, LLMs may be used privately to answer questions, analyse, review and suggest, but not to create. LLM-written comments, issue bodies, PR descriptions and docs are banned; trivial changes, machine translation and verified LLM-found bugs need disclosure, and LLM-created code is only allowed in a pre-arranged, disclosed experiment.
- [scikit-image](https://areweaifriendly.com/projects/scikit-image.md): Not AI-friendly. AI may not be used to create novel code such as algorithms or tests, or to write PR descriptions and comments. It is allowed for exploring the codebase, iterating on hand-written code, trivial fixes and mechanical changes, with AI use declared and AI commits tagged Assisted-by.
- [scikit-learn](https://areweaifriendly.com/projects/scikit-learn.md): AI-friendly with conditions. AI-assisted code is accepted if the contributor has reviewed, understood and tested it and states AI use in the PR description. Fully automated issues and PRs are not accepted, and AI-generated text in descriptions and comments is not wanted beyond grammar help.
- [SciPy](https://areweaifriendly.com/projects/scipy.md): AI-friendly with conditions. AI-assisted pull requests are accepted if AI use is disclosed in detail and the author understands and can explain the code. Autonomous agents are not allowed, and AI may not speak for contributors in issues or discussions except for translation or grammar fixes.
- [SDL](https://areweaifriendly.com/projects/sdl.md): Not AI-friendly. Generative AI should not be used in any way when contributing to SDL. Every pull request confirms that it contains no LLM-generated code, and AI-generated comments and code are not wanted in bug reports or pull requests.
- [SearXNG](https://areweaifriendly.com/projects/searxng.md): AI-friendly with conditions. All AI use must be disclosed with the tool and extent, the contributor must fully understand the code, and AI may never be the main author of a PR. Issues and PR descriptions must be fully human-written, with AI translation of self-written text allowed.
- [Servo](https://areweaifriendly.com/projects/servo.md): Not AI-friendly. Content generated by LLMs is not allowed in code, documentation, pull requests, issues or comments. AI may be used to translate into English, to understand the codebase, or to help find bugs, as long as findings are verified before filing.
- [Sphinx](https://areweaifriendly.com/projects/sphinx.md): AI-friendly with conditions. AI-assisted pull requests are accepted if AI use is disclosed in detail and the author understands and can explain the code. Autonomous agents are not allowed, and AI may not write descriptions, comments or issues except for translation or grammar fixes.
- [stb](https://areweaifriendly.com/projects/stb.md): Not AI-friendly. Code or documentation generated by LLMs or other generative AI is not accepted, and AI autocompletion should not be used either. Pull requests and issues must be written by the contributor, even in imperfect English.
- [SymPy](https://areweaifriendly.com/projects/sympy.md): AI-friendly with conditions. AI-assisted patches are accepted if substantial AI use is disclosed and the author understands the code. AI may not write descriptions or speak for contributors except for translation or grammar, and mostly AI-generated work from new contributors will likely be closed.
- [Tusky](https://areweaifriendly.com/projects/tusky.md): Not AI-friendly. Tusky does not want any contributions, code or otherwise, created with generative AI tools.
- [Twisted](https://areweaifriendly.com/projects/twisted.md): Not AI-friendly. Generative AI output is not allowed in code, documentation, issues, PR descriptions or discussions, and contributors self-certify that they used none. Findings learned from an LLM, such as security issues, are accepted if fully reviewed and written up by hand.
- [typescript-eslint](https://areweaifriendly.com/projects/typescript-eslint.md): AI-friendly with conditions. AI-assisted contributions are not banned, but AI-generated code must be reviewed closely and only used for work the contributor could handle without AI. AI-generated issues, comments and PR descriptions should be avoided in favour of the contributor's own words.
- [uv](https://areweaifriendly.com/projects/uv.md): AI-friendly with conditions. AI is supported as a coding tool, with contributors responsible for what they submit and a human who understands the work in the loop. Autonomous agents are not allowed. Issues, PR bodies and replies must be in the contributor's own words; quoted AI context must be marked and explained, and AI editing for non-native speakers is accepted.
- [Veloren](https://areweaifriendly.com/projects/veloren.md): Not AI-friendly. Content generated by LLMs such as Copilot or ChatGPT is presumed to be tainted and must not be committed. The same applies to AI-generated textures, models, sound effects and music.
- [Vim](https://areweaifriendly.com/projects/vim.md): AI-friendly with conditions. AI may be used for contributions if its use is disclosed. AI-generated code must follow the Vim code style and be properly tested.
- [Wagtail](https://areweaifriendly.com/projects/wagtail.md): AI-friendly with conditions. AI assistance is permitted but purely AI-generated contributions are not; contributors must understand the solution and fill in the AI usage part of the PR description. PR descriptions and replies to review must be written by the contributor, with AI help for English allowed.
- [WordPress](https://areweaifriendly.com/projects/wordpress.md): AI-friendly with conditions. AI tools may assist with code, tests, docs and issues. Contributors must understand and review every line, keep everything GPL-compatible, and disclose meaningful AI help in the PR description or Trac ticket. AI may draft issue and PR descriptions if the details are reviewed and verified.
- [Xarray](https://areweaifriendly.com/projects/xarray.md): AI-friendly with conditions. AI-generated code is accepted if the submitter has read, tested and fully understands every change; large AI-assisted changes should be discussed in an issue first. PR descriptions, issue comments and review replies must be in the contributor's own words, with AI allowed only to polish language.
- [yt-dlp](https://areweaifriendly.com/projects/yt-dlp.md): Not AI-friendly. LLMs, agents and other AI tools are strictly forbidden for any kind of contribution: issues, bug reports, pull requests, PR descriptions, comments and translation. Violators may be blocked without warning.
- [Zulip](https://areweaifriendly.com/projects/zulip.md): AI-friendly with conditions. Any tools including AI may be used to understand the code and write it, but contributors must understand, test and be able to explain every change. AI-generated messages are not allowed in the development community, apart from help with spelling, grammar or translation, and LLMs should not be pointed at the code to find problems.
- [CPython](https://areweaifriendly.com/projects/cpython.md): AI-friendly. AI tools are fine for code, issues and comments as long as the author reviews the output and takes responsibility. Disclosing AI use is appreciated but not required.
- [curl](https://areweaifriendly.com/projects/curl.md): AI-friendly with conditions. Code written with AI help is accepted if it meets the normal standards. AI-found bug and security reports must say so and be verified by a human, and pasted AI-generated reports and explanations are discouraged.
- [elementary OS](https://areweaifriendly.com/projects/elementary.md): Not AI-friendly. Contributions generated by LLMs and chatbots are banned everywhere, including code, docs, issues and artwork. Machine translation of issue text into English is the only exception.
- [Fedora](https://areweaifriendly.com/projects/fedora.md): AI-friendly with conditions. AI-assisted contributions are allowed. Contributors are fully accountable for them and must disclose AI use when a significant part comes from a tool unchanged.
- [Gentoo](https://areweaifriendly.com/projects/gentoo.md): Not AI-friendly. Contributing any content created with the help of AI language tools is forbidden, citing copyright, quality and ethical concerns.
- [Ghostty](https://areweaifriendly.com/projects/ghostty.md): AI-friendly with conditions. AI help is welcome for code, issues and discussions, but every use must be disclosed and a human must fully understand and edit the result. AI-generated media is not allowed.
- [Linux kernel](https://areweaifriendly.com/projects/linux-kernel.md): AI-friendly with conditions. AI-assisted patches are accepted when the AI use is tagged with Assisted-by and a human reviews the work, signs off and takes full responsibility. AI agents may not sign off or send anything on their own.
- [NetBSD](https://areweaifriendly.com/projects/netbsd.md): Not AI-friendly. Code generated by LLMs is treated as tainted and may not be committed without prior written approval from the core team.
- [QEMU](https://areweaifriendly.com/projects/qemu.md): Not AI-friendly. Contributions that include or derive from AI-generated content are declined over licensing concerns. Using AI for research, static analysis or debugging is fine if its output stays out of the patch.
- [Zig](https://areweaifriendly.com/projects/zig.md): Not AI-friendly. A strict no-LLM policy: no generated code or prose, no AI editing or translation, and no using LLMs to find bugs.

## Data

- [All projects as JSON](https://areweaifriendly.com/projects.json): full structured data, including quotes and sources
- [Everything in one file](https://areweaifriendly.com/llms-full.txt): every project page as Markdown

## Optional

- [Methodology](https://areweaifriendly.com/about/): how ratings are made and checked