writing-dna-skill Writing Distiller — Extracting Reusable Writing DNA from 20 Articles

Skill Overview

writing-dna-skill is a writing distillation tool: it extracts a writer, account, or brand’s writing DNA from at least 20 complete historical articles. This is not content summarization, but an actionable set of reusable rules covering language, structure, topic selection, source material, cognitive frameworks, and visual style. The output is a Writing-DNA.md that can directly drive AI writing.

Use Cases

1. Turning an account or author’s style into an asset: Operators of public WeChat accounts, blogs, or newsletters can consolidate the writing style scattered across their historical articles into a set of transferable and reusable style rules, ensuring consistency when handing writing responsibilities to someone else or migrating to another platform.
2. AI style-replication writing: Produce new articles in a style resembling that of the author distilled from the DNA—the wording and tone, article structure, image rhythm, and layout density can all be similar. This is suitable for establishing a style baseline before producing content at scale.
3. Style comparison and team-wide consistency: Compare how different authors express themselves on the same topic, or create a unified style guide for a content team or branded publication, ensuring that writing by multiple people reads as if it came from the same pen.

Core Features

1. Six-layer distillation system (L1–L6): The system analyzes everything from surface-level language—word frequency, sentence length, and punctuation—to article structure, including hook types, body architecture, and ending styles. It then examines topic-selection logic, source-material strategy, cognitive frameworks, and visual style, including image strategy, layout hierarchy, and color usage. Each layer produces an independent document. Images are also opened and examined, with screenshots and charts incorporated into the chain of argument as part of the analysis.
2. Generating an integrated Writing-DNA.md: The six layers of analysis are compressed into a single rule document of no more than 4,000 Chinese characters. Anyone—or any AI—who has never read the original author’s articles can write an article in a roughly similar style scoring above 70 out of 100 after reading it. Chinese and English bilingual outputs are supported.
3. A consistency workflow for writing according to the DNA: Whenever the distillation output is used to write a new article, the process requires first reading all four layer-specific outputs together with the integrated document, then reading five original articles closest in genre and subject matter to calibrate the voice. After writing, the draft is cleaned according to anti-AI-writing rules to prevent style drift—avoiding the result of “sounding similar, but not actually being similar.”

Frequently Asked Questions

How many articles are needed to distill writing DNA?

At least 20 complete articles. Articles should be in .md or .txt format. It is also recommended that they cover different periods and types—such as in-depth analyses, short commentaries, and interviews—so that the extracted rules are stable and are not skewed by the incidental writing habits of a few articles.

How is this different from simply pasting a few articles and asking AI to imitate them?

Direct imitation requires pasting the articles again each time, and AI often learns only superficial wording and phrasing. The writing distiller produces a layered, long-term reusable set of rules covering not only “how to write,” but also “how to think,” including topic angles, criteria for selecting source material, and core assumptions about a subject, as well as “how to present,” including imagery and layout. Once distilled, every subsequent article can be written according to the same DNA, resulting in significantly greater consistency.

Does it support English articles? How similar can the resulting writing be?

Yes. The skill provides separate output templates for Chinese and English corpora and can also process mixed Chinese-English material. The degree of similarity depends on the quality of the source material. Following the standard distillation process, the goal is for a writer—human or AI—who has never read the original author’s work to produce an article with a roughly similar style scoring above 70 out of 100. The skill also explicitly requires replicating the writing method rather than the content; the original articles’ opinions are not carried over into new articles.