human-writing
An open-source writing skill for Chinese creators and editors that uses material and narrative constraints to produce more human-sounding drafts.
Tool overview
Adoption verdict: it is worth a small, hands-on trial, but the current discussion does not justify treating it as a mature, broadly validated writing product. There are five evidence records, including three Zhihu articles and one X mention; roughly 45 Zhihu upvotes and three comments show attention. However, the only GitHub record points to gethuman-sh/human, not the target repository, so it cannot prove human-writing’s stars, growth, or engineering maturity. The evidence therefore proves interest more than usefulness: discussion is relatively rich in rule explanations and opinions, while reproducible tests, comparative samples, and long-term feedback remain limited.
In practice, it is better understood as a Chinese writing specification and revision workflow for an LLM, not as a standalone writing model. The articles describe a requirement for at least five sourced, concrete materials before producing a long piece on a real-world topic, together with checks on who is speaking, how they know, why they are writing now, and what the material can support.