Report · github.com
MicroZig to Adopt Strict No LLM / No AI Policy

github.com
Source published October 29, 2026
Summary: 1 min read
44/100
Summary
The Zig Embedded Group has adopted a Code of Conduct for its GitHub organization and Discord server that takes effect October 28, 2026. The policy, largely taken from Zig’s Code of Conduct, limits these spaces to focused technical work on MicroZig and Zig Embedded Group projects and requires participants to help maintain a positive environment during disagreements. The rules prohibit all LLM-generated code and prose, paraphrasing of LLM output, and LLM use for editing, translation, brainstorming shared in project spaces, or bug finding. Discussion of chatbot or LLM services is also prohibited. English is encouraged but not required, and participants may post in their native language for others to translate using tools of their choice. The policy calls for inclusive language, respect, constructive criticism, empathy, and appreciation of others’ work. It bars harassment, personal attacks, off-topic discussion, disclosure of private information, public discussion of the Code of Conduct, accusations of violations, exclusionary comments, and destructive behavior. Matthew Knight and Felix Queißner handle reports and say they will swiftly remove antagonistic or destructive participants. The policy also covers private harassment outside the organization’s official platforms, with offenders subject to being blocked from participation.
Key Takeaways
- The Zig Embedded Group’s Code of Conduct applies to its GitHub organization and Discord server beginning October 28, 2026.
- MicroZig and Zig Embedded Group spaces prohibit LLM-generated or LLM-paraphrased content, as well as LLM use for editing, translation, brainstorming, and bug finding.
- The policy prohibits discussing chatbot services, the Code of Conduct, or alleged violations in the project’s GitHub and Discord spaces.
- Matthew Knight and Felix Queißner receive conduct reports, and the policy permits blocking participants for private harassment that occurred outside official project platforms.
What the discussion said
The thread spent less time on MicroZig itself than on whether an AI-free contribution policy is a legitimate defense of programming craft or an impractical refusal to use useful tools. Supporters saw a small embedded-systems community choosing a deliberate working culture: contributors should understand every bug report, design idea, and patch well enough to explain it without an opaque model session. They argued that this filters out drive-by agent submissions, preserves distinctive human judgment, and lets maintainers reject low-effort AI output without litigating every pull request. Several also respected the maintainer’s candid framing: incompatible preferences require a choice, and people who want AI-assisted development can fork. Critics focused on the unusually broad prohibition, especially barring models from bug discovery and private brainstorming. They argued that a manually verified report remains valuable regardless of how the lead was found, and that refusing AI-assisted security analysis could leave defects undiscovered. The rule’s enforceability was another fault line: some considered provenance impossible to prove once output is rewritten, while others compared it to contributor attestations and said enforcement can target suspicious patterns rather than perfect detection. A middle view emerged that the real standard should be comprehension, not total abstinence: tools may surface ideas, but contributors must independently grasp the behavior, consequences, and proposed fix. The policy’s tone earned more respect than similar bans, even among readers unconvinced by its scope.
Where opinion split
The sharpest dispute is whether banning LLMs from bug finding and brainstorming protects maintainable, human-understood work or simply suppresses useful, verified technical input. Defenders say a project can prioritize contributors who can reason independently and avoid the review burden and homogenization created by agent-driven submissions. Opponents say the origin of a lead is irrelevant once a human validates and explains it, and that excluding AI-assisted vulnerability discovery is a needless safety risk.
Coverage & discussions
Hacker News · github.com
Community discussions
Community Sentiment
Positives
- A strict AI-free lane gives maintainers a clean way to reject mass-produced agent pull requests before review bandwidth is swallowed by polished-looking noise.
- Requiring contributors to explain bugs and fixes from their own understanding can keep embedded code auditable instead of making maintainers reconstruct an unseen model conversation.
- Commenters appreciated that the policy treats AI-assisted development as a different workflow rather than a moral failing, leaving forks open for people who want it.
- Several readers see the ban as a way to resist the uniform style and shallow reasoning they associate with LLM-generated documentation, designs, and code.
Concerns
- Barring models from finding bugs looks self-defeating to critics: a validated defect does not become less real because AI helped expose it.
- The policy reaches into private brainstorming and rewritten prose, a scope critics consider impossible to verify and therefore more performative than operational.
- Some warn that rejecting AI-assisted security investigation could leave exploitable flaws undiscovered, particularly where software is exposed beyond a low-stakes device.
- Skeptics argue that capable contributors can disguise model involvement anyway, so detection will punish obvious users rather than reliably preserve human-only work.
Related Articles

About LLMs at Zig Days
May 28, 2026

fx :Tiny, open, native coding agent.
Aug 18, 2026

Pledging another $400k to the Zig software foundation
Jun 22, 2026

OpenHeronBrowser – a new browser engine in Rust, built with Claude. Acid3 97/100, runs Discord, VS Code for the Web, and Windows 1.01 in v86
Oct 27, 2026

R/open source bans all solo projects and projects using LLMs
Oct 25, 2026