An open-source contributor fixed bugs across a multi-language game-dev toolchain using LLMs in 2026. The community they contribute to explicitly opposes AI. They haven’t told anyone.
The developer had filed PRs against the toolchain before but found it too complicated — tools written in multiple programming languages, layers they couldn’t wrap their head around. LLMs removed that barrier:
- Language switching — LLMs move between the toolchain’s multiple programming languages without the context-switching cost that stopped this contributor before
- Debug tracing — LLMs add trace statements to follow a bug’s flow through each layer
- Test generation — LLMs write automated tests that reproduce the bug
- Bug fixes — LLMs produce correct fixes once the bug is traced and reproduced
The HN post cuts off mid-sentence at the dilemma: how to convince the team to merge a PR containing entirely LLM-generated code when the community culture forbids it. The question is unresolved. But the pattern is recognizable — developers in AI-skeptical communities are quietly using LLMs to contribute, then handling the social problem separately.
NOTE
The poster isn’t hiding AI use to deceive reviewers about code quality. They’re hiding it because the community’s norms make AI-generated PRs unmergeable on principle — regardless of whether the code is correct.
The bottleneck isn’t whether LLMs can fix bugs in complex codebases — it’s whether communities will accept fixes from contributors who used them.