Review Taiwan Traditional Chinese drafts with speak-human-tw
Use speak-human-tw when an agent should identify AI-writing fingerprints in Taiwan Traditional Chinese drafts, suggest localized revisions, and preserve the writer's intent instead of generating generic copy.
npx skills add agentskillexchange/skills --skill review-taiwan-traditional-chinese-drafts-with-speak-human-tw
speak-human-tw is a Traditional Chinese editing skill for Claude Code, Codex, Cursor, and other Markdown-reading agents. Use it when an operator has an existing Taiwan Traditional Chinese draft and wants an agent to find AI-sounding patterns, China-market wording, half-width punctuation, generic transitions, and other synthetic writing signals before final review. Invoke this instead of using a normal writing product when the workflow is draft critique and constrained revision: provide the existing text, choose annotation-only or direct rewrite mode, review the flagged issues, and accept only the edits that preserve the author’s voice and local usage. The scope boundary is Taiwan Traditional Chinese AI-fingerprint review and localization, not a universal content writer, generic translation tool, or broad humanizer card.
What this skill actually does
Inputs and prerequisites: Claude Code, Codex, Cursor, or another agent/runtime that can read a SKILL.md file; Taiwan Traditional Chinese draft text.
Setup notes: For Claude Code, clone the repository into ~/.claude/skills/speak-human-tw. For Codex one-shot usage, clone the repository, run codex exec -C . with instructions to read ./SKILL.md, and paste the draft text. Cursor and symlink install options are documented in the upstream README.
Source and verification boundary: use https://github.com/Raymondhou0917/speak-human-tw as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Multi-Framework workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.