菜单
This skill should be used when the user asks to "write in a human tone", "avoid AI style", "make this sound human", "remove AI patterns", "write naturally", or when asked to write blog posts, articles, emails, social media posts, or any text where sounding human is important.
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。
基于 SOC 职业分类
Create an Obsidian note documenting a non-code work decision — research, investigation, planning, or design choices. Entirely interview-based with no git context required. Use when the user wants to log a problem they investigated, options they considered, and a decision they reached.
Create a retrospective Obsidian note from a development task. Given a task name and brief background, the AI collects git and code context, interviews the user about decisions and insights, and writes a structured markdown log to the vault. Use when the user wants to document a completed task, feature, or bugfix as a retrospective note.
Create a team-facing work log from a development task and push it directly to a shared GitHub repository via gh CLI. Collects git context automatically and asks 2–3 focused questions for weekly team review. Use when the user wants to share what they worked on with their team.
| name | human-writer |
| description | This skill should be used when the user asks to "write in a human tone", "avoid AI style", "make this sound human", "remove AI patterns", "write naturally", or when asked to write blog posts, articles, emails, social media posts, or any text where sounding human is important. |
| version | 1.0.0 |
Write in a natural human style. Every rule below is active for the duration of this task.
Plain text only. No markdown formatting of any kind:
Use paragraph breaks to organize ideas. If you need to list things, write them into a sentence with commas or "and".
No special characters that aren't on a standard US keyboard. Specifically banned:
If you feel the urge to use an em dash, stop and rewrite the sentence.
Avoid all AI-clichéd words and phrases. See references/ai-tropes.md for the full list.
Short version: don't use words like delve, leverage, robust, seamlessly, tapestry, ecosystem, journey, navigate, realm, invaluable, harness, crucial, vital, groundbreaking, game-changer, meticulous, elevate, or unlock. Don't open with "In today's world" or "It's worth noting that". Don't close with a summary that restates everything you just said.
No false profundity. This means no "It's not X — it's Y" constructions.
No self-answered questions. Don't write "The result? Devastating." or "Why does this matter? Because..." and similar.
No rule-of-three padding. If you catch yourself writing three parallel phrases for emphasis ("faster, smarter, and more efficient"), pick one or find a real reason to list them.
No fractal summaries. Don't restate the point you just made at the end of a paragraph or at the end of the piece.
No anaphora for effect. Don't start three sentences in a row with the same word to create dramatic rhythm.
No dramatic one-sentence fragments used as paragraphs for emphasis.
Vary sentence length. Short sentences and long sentences should mix naturally. Reading a paragraph where every sentence is roughly the same length feels robotic even if the words are fine.
Write as if talking to someone, not presenting to them. Imagine explaining this to a person sitting across from you, not delivering a keynote.
Don't hedge everything. "It's important to note that" and "it's worth mentioning" are filler. Just say the thing.
Don't over-qualify. One caveat is enough. Not every claim needs three qualifiers.
Good writing has a point of view. It makes choices. It says something direct and moves on. It doesn't pad. It doesn't perform expertise — it just has it.
If a sentence isn't doing work, cut it.
