菜单
Add or run Autor Ventura structural editorial evals for routing publication packs voice and factual gates. Use when changing prompts router logic memory behavior or review constraints. Do not use when judging literary quality from one generated draft without a defined test contract.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
基于 SOC 职业分类
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/venturalabs-ai/ventura.autor --skill editorial-eval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。
正在显示 SKILL.md
Detect and remove AI-sounding patterns in Brazilian Portuguese text while preserving meaning tone and register. Use when finalizing or reviewing PT-BR copy drafts posts or chapters that must not read like generic AI output. Do not use when the text is already approved by a human editor or when the task is only factual verification without style review.
Verify foreign-language sentences in a text for correctness of spelling diacritics and register. Use when a text contains phrases that are not in the main language or when characters places titles or terms in another language must be written correctly. Do not use when the whole text is in one language and there is nothing to verify.
Update Autor Ventura voice memory from approved author samples preferences rejections and revision history without inventing personal traits. Use when recurring writing should preserve a real author's established voice. Do not use when no approved samples or preferences exist or when the task needs only one-off tone adaptation.
| name | editorial-eval |
| description | Add or run Autor Ventura structural editorial evals for routing publication packs voice and factual gates. Use when changing prompts router logic memory behavior or review constraints. Do not use when judging literary quality from one generated draft without a defined test contract. |
scripts/ evals/ and editorial-evals/.python scripts/run_evals.py and the repository's documented eval commands.