dongbeixiaohuo/writing-agent

🚀 一个基于 Claude Code (Skills + Subagents) 的“去AI味”全栈写作系统。不仅防套路,更通过专属规则强制注入人类观点与细节,搭配读者测试评估与自动图文排版。全面支持 DeepSeek / 智谱GLM / MiniMax 等国产低成本大模型,提供从选题、风格建模到审稿发布的高维全自动写作工作流。

392Stars on the repository
26Mods indexed here, across every type
9d agoLast push, which is what freshness is scored on
MITLicence, which decides whether bodies are shown

article-illustrator

01

dongbeixiaohuo/writing-agent

Agent Claude Code

An article-illustration planning agent that designs cover images and in-text illustrations for a finished article and its publishing platform.

not rated 392 9d ago A 39 tokens original MIT

concretizer

02

dongbeixiaohuo/writing-agent

Agent Claude Code

A writing agent that turns abstract ideas into comparisons, visible scenes, and concrete actions.

not rated 392 9d ago A 41 tokens original MIT

edit-diff-learner

03

dongbeixiaohuo/writing-agent

Agent Claude Code

A writing review agent that compares an AI first draft with the user-approved final version to extract reusable lessons.

not rated 392 9d ago A 73 tokens original MIT

editor-review

04

dongbeixiaohuo/writing-agent

Agent Claude Code

An editorial review agent that checks an article’s structure, information density, writing style, and signs of formulaic AI wording.

not rated 392 9d ago B 40 tokens original MIT

empathy-designer

05

dongbeixiaohuo/writing-agent

Agent Claude Code

A planning agent for making articles more relatable and more likely to be shared, using real details from the topic and outline.

not rated 392 9d ago A 53 tokens original MIT

fact-checker

06

dongbeixiaohuo/writing-agent

Agent Claude Code

A fact-checking subagent that reviews a final article against its evidence ledger and, where needed, external sources. An evidence ledger is a record of the sources supporting factual claims.

not rated 392 9d ago A 70 tokens original MIT

html-exporter

07

dongbeixiaohuo/writing-agent

Agent Claude Code

A tool that turns a finished Markdown article into an HTML file using a selected visual layout. HTML is the web-page format used by browsers.

not rated 392 9d ago A 36 tokens original MIT

humanizer

08

dongbeixiaohuo/writing-agent

Agent Claude Code

A text-editing agent that removes common signs of AI-written prose while preserving the original facts, experiences, evidence, viewpoints, and style.

not rated 392 9d ago A 65 tokens original MIT

memory-loader

09

dongbeixiaohuo/writing-agent

Agent Claude Code

A writing-memory loader that gathers lessons from earlier article revisions and publishing reviews into a short set of reusable preferences.

not rated 392 9d ago A 79 tokens original MIT

opening-tournament

10

dongbeixiaohuo/writing-agent

Agent Claude Code

A writing subagent that creates three very different opening paragraphs for an article, based on earlier planning and evidence files.

not rated 392 9d ago A 62 tokens original MIT

outline-architect

11

dongbeixiaohuo/writing-agent

Agent Claude Code

An outline-planning agent that turns a topic, viewpoint, source material, and style requirements into an article structure.

not rated 392 9d ago B 31 tokens original MIT

performance-review

12

dongbeixiaohuo/writing-agent

Agent Claude Code

A post-publication review process that compares an article’s real performance data with its title, body, cover, and distribution context.

not rated 392 9d ago A 64 tokens original MIT

position-engine

13

dongbeixiaohuo/writing-agent

Agent Claude Code

A writing-planning subagent that gives an article a clear, strongly argued position after its main topic has been chosen.

not rated 392 9d ago B 59 tokens original MIT

pre-publish-review

14

dongbeixiaohuo/writing-agent

Agent Claude Code

A pre-publication review process for articles that checks whether the content offers readers value and whether publishing it creates avoidable risks.

not rated 392 9d ago A 29 tokens original MIT

research-expert

15

dongbeixiaohuo/writing-agent

Agent Claude Code

A research agent that digs up highly specific human details, costs, and industry practices for an article, while keeping a ledger of factual sources.

not rated 392 9d ago B 67 tokens original MIT

title-designer

16

dongbeixiaohuo/writing-agent

Agent Claude Code

An agent that designs article headlines using predefined headline patterns and factual limits. It is called by a larger writing workflow at a specific stage.

not rated 392 9d ago A 44 tokens original MIT

topic-generator

17

dongbeixiaohuo/writing-agent

Agent Claude Code

A writing assistant that creates possible article topics from current events, the writer’s previous work, and competitor analysis.

not rated 392 9d ago A 47 tokens original MIT

topic-research

18

dongbeixiaohuo/writing-agent

Agent Claude Code

A research subagent for choosing article topics. It checks whether a topic is worth writing about by examining trends, popular articles, reader concerns, and available material.

not rated 392 9d ago A 49 tokens original MIT

wechat-reader-test

19

dongbeixiaohuo/writing-agent

Agent Claude Code

A pre-publication reader test that checks how readers may react to an article on WeChat, Toutiao, or Zhihu. It uses text-based judgments rather than real performance data.

not rated 392 9d ago A 81 tokens original MIT

writing-clarifier

20

dongbeixiaohuo/writing-agent

Agent Claude Code

A writing-requirements clarification agent that gathers the topic, audience, style, length, examples, and publishing platform before writing begins.

not rated 392 9d ago B 50 tokens original MIT

writing-executor

21

dongbeixiaohuo/writing-agent

Agent Claude Code

A Chinese-language writing subagent that turns prepared research and planning files into a first draft. It supports a light workflow and a collaborative workflow, each with required input checks.

not rated 392 9d ago B 37 tokens original MIT

At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: