Back to directory

fluent-korean (Claude Code output-style)

A Claude Code output-style plugin that stops Claude from writing broken machine-like Korean, correcting dropped particles, telegraphic phrasing, and mismatched vocabulary in favor of clear, natural sentences.

fluent-korean (Claude Code output-style)

fluent-korean is an output-style plugin that instructs an LLM, particularly Claude and coding agents, to write clear, fluent Korean. It prioritizes unambiguous meaning over elegant prose, aiming for reasonably natural sentences rather than literary style, and is most useful when a prompt is written in Korean or when the agent's output needs to include Korean text.

It ships as two variants, one that keeps Claude Code's own coding instructions alongside the Korean writing guidance and one without them, and the guideline text itself can also be pasted into other AI environments as a plain writing instruction. It is aimed at correcting quality loss that compounds when models or agents exchange Korean text with reduced token budgets, including in multi-agent setups.

The Korean portions of the README are written by the author, a Korean-literature major, rather than machine-translated.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.