用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/XucroYuri/how-to-make-script --skill voice-style-calibration命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | voice-style-calibration |
| description | 需要为角色、IP、品牌或交互特定写作提供语音风格指南时使用。 |
Use this skill when the user asks for language style calibration, character voice separation, IP continuity, brand tone control, or stronger lived-in expression before drafting.
voice_style_guide: a calibration guide with voice anchors, lived-pressure notes, register envelope, variability budget, and drift warnings.When certainty = lost:
Ask: "Say one line this character would say that no other character in this story would say." One authentic line is more useful than a voice taxonomy. Let it anchor the calibration.
When source = discover:
Surface the voice from character psychology: "What does this person reach for when they are under pressure — precision, deflection, humour, silence?" Register follows from need, not from aesthetics.
When source = construct:
Apply the full workflow: speaking identity → expression job → voice anchors → lived-pressure notes → register envelope → variability budget → drift warnings. Sample lines are demonstrations, not canon.
When focus = language:
Primary activation context. Load ka.character-voice-consistency, ka.verbal-rhythm, and ka.embodied-text-pressure first. Register envelope and variability budget are the most actionable outputs.
When focus = character:
Voice is character psychology made audible. Before calibrating register, confirm the character's wound-lie-want-need axis. Voice that contradicts that axis is not authentic voice, even if it sounds interesting.
wp.voice-style-guiderb.voice-style-guideka.character-voice-consistencyka.cross-protocol-referral-edgeska.dialogue-subtextka.embodied-text-pressureka.ip-voice-continuityka.register-adaptationka.tone-writing-moves