MicroZig bans LLM help, even for translation and finding bugs
The Zig Embedded Group's new contribution rules reach beyond generated code to proofreading, brainstorming and chatbot-assisted issue comments.
MicroZig has adopted a strict no-LLM policy that reaches well beyond refusing generated code. Its new Code of Conduct rules out using language models for spelling corrections, translation, brainstorming that feeds a contribution, and finding bugs. For developers accustomed to treating those tasks as routine assistance, the policy draws a much wider boundary around participation.
The document says the policy takes effect October 28, 2026. GitHub records the adoption commit at 00:13:46 UTC on October 29, with a further edit at 00:37:14 UTC. Those are distinct dates: the effective date comes from the policy itself, while the repository timestamps establish when the text was committed.
The restriction includes the words around the code
MicroZig is a toolbox for building embedded applications in Zig. The Code of Conduct applies to the Zig Embedded Group GitHub organization and Discord server. It prohibits LLM-generated code or prose, paraphrasing generated content, and using an LLM to edit text. It also explicitly says a comment is unwelcome if a chatbot supplied advice for it, even when the contributor writes the final prose.
The translation rule comes with a concrete alternative: English is encouraged but not required, and contributors may post in their native language. The document says readers can use their own translation tools to interpret those words. This is a rule for the named project spaces; its stated scope does not establish a ban across every project using Zig.
A clear rule, disputed implications
An October 29 Hacker News discussion shows disagreement over the breadth of the restrictions. Participants questioned the prohibition on finding bugs with an LLM and debated whether private brainstorming can be meaningfully enforced. Other commenters defended expectations that contributors understand and take responsibility for what they submit. These are individual reactions, not evidence of a consensus among MicroZig contributors.
The published policy establishes the participation rules. It does not demonstrate that LLM assistance always harms code quality, nor does it describe a technical system for detecting private chatbot use. The immediate change is practical: contributors now have an explicit restriction covering both submitted artifacts and several kinds of assistance used to prepare them.