Dan Koe Writer

v1.0.1

Dan Koe 写作方法论工具箱。基于 OpenClaw Agent 模型,把爆款内容拆解成「创意积木」,再组装成新文章。触发场景:(1) 用户要拆解爆款文章并提取写作要素;(2) 用户需要写作方向或创意灵感;(3) 用户要生成完整的公众号/小红书/推特文章;(4) 用户想学习 Dan Koe 的 APAG 框架...

0· 131·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for alleliu/dan-koe-writer.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "Dan Koe Writer" (alleliu/dan-koe-writer) from ClawHub.
Skill page: https://clawhub.ai/alleliu/dan-koe-writer
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install dan-koe-writer

ClawHub CLI

Package manager switcher

npx clawhub@latest install dan-koe-writer
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
Purpose & Capability
The name/description (Dan Koe writing toolbox) matches the included scripts and reference markdown: goldmine.py extracts building blocks from articles, spark.py composes ideas from the local knowledge base, and write.py assembles prompts for generating articles. The ability to fetch WeChat pages (playwright) and read local article files is coherent for a tool that 'tears down' viral posts and constructs new ones.
Instruction Scope
SKILL.md and the scripts instruct the agent to run the included scripts and then call the agent's own model on the generated prompts. This stays within writing/analysis scope. Two things to be aware of: (1) goldmine.py and the scripts can read a user-supplied file path (--file) anywhere the agent or user points to, so if an agent is allowed to pick arbitrary paths it could read arbitrary local files; (2) fetching remote pages (Playwright) causes network access to the given URL. Both behaviors are expected for this skill but users should avoid passing sensitive file paths or secrets as input.
Install Mechanism
No install spec is declared (instruction-only), which is low risk. The code depends on third-party tooling (playwright) that is not installed automatically; the SKILL.md and scripts print guidance about installing playwright but there's no automated installer. This is a usability/incoherence note (missing declared dependency), not an indication of malicious activity. Playwright will download browser binaries when installed, which is normal but worth noting.
Credentials
The skill requests no environment variables or credentials. The scripts operate on local files (references/knowledge and user-provided inputs) and do not attempt to read environment secrets or cloud credentials.
Persistence & Privilege
always:false and no special platform privileges. The skill will read and may append to its own knowledge files (references/knowledge) when the user runs the 'load' flow, which is expected. It does not modify other skills or system-wide agent settings. Allowing the agent to choose file paths when invoking the scripts would increase risk of unwanted file reads/writes; restrict inputs if that is a concern.
Assessment
This skill appears to do what it says: analyze articles, build a local 'creative blocks' library, and produce prompts/contexts for an LLM to generate posts. Before installing or invoking it: 1) Be aware the scripts can fetch web pages (network access) and read any file paths you pass in — do not supply sensitive files or credentials. 2) The code uses Playwright (not auto-installed) which will download browser binaries if you install it; install third-party libs from trusted sources and review them. 3) The skill writes to its own references/knowledge folder when you save extracted blocks — if you prefer no on-disk persistence, avoid running the save/load operations. 4) The skill relies on the agent’s model to perform LLM extraction/generation; ensure your agent policies and model access controls align with how you want the AI to run. If you want extra caution, run the scripts in an isolated environment or container and review any extracted prompts/outputs before sending them to an external model.

Like a lobster shell, security has layers — review code before you run it.

latestvk97d9kavx1p7arc426vhq5g89983vtm0
131downloads
0stars
2versions
Updated 4w ago
v1.0.1
MIT-0

⚠️ 免责声明:本技能基于 Dan Koe 公开分享的写作方法论整理,用于帮助用户学习并应用这些方法论创作原创内容。所有示例积木仅供学习参考,用户应使用自己的原创文章进行拆解和写作。

Dan Koe Writer

基于 Dan Koe 写作方法论的多智能体写作工具,把爆款内容拆解成「创意积木」,再组装成新内容。

注意:本工具调用 OpenClaw Agent 的模型(minimax-m2.7),不需要额外配置 API Key。

核心流程

爆款内容
    ↓ [goldmine.py]
积木库(7类积木)
    ↓ [spark.py]
创意方向(供 Agent 使用)
    ↓ [Agent 调用自己的模型]
完整文章

工作目录

dan-koe-writer/
├── scripts/
│   ├── goldmine.py   # 拆解爆款 → 提取积木(输出 prompt 供 Agent 调用)
│   ├── spark.py      # 从积木库生成创意方向(支持 JSON 输出)
│   └── write.py      # 从积木库构建写作 context(输出 prompt 供 Agent)
└── references/
    └── knowledge/    # 积木库(7类 Markdown 文件)

积木库初始有 Dan Koe 方法论示例积木,可直接使用。

积木库说明

文件内容使用场景
hooks.mdAttention 钩子标题/开头抓住注意力
paradoxes.md悖论引发认知冲突
arcs.md转化弧故事线:Before→顿悟→After
core_problems.md核心问题引发读者共鸣
golden_phrases.md金句收尾、被转发引用
structures.md结构模式APAG/PAS/Before-After
perspectives.md观点POV(独特视角)

使用方式

1. 拆解爆款内容

cd <skill>/scripts

# 抓取微信文章
python goldmine.py "https://mp.weixin.qq.com/s/xxxxx"

# 从本地文件拆解
python goldmine.py --file article.txt --source "文章标题"

# 直接输入文本
python goldmine.py --text "文章内容..." --source "手动输入"

输出:告诉 Agent 需要调用自己的模型提取积木。

2. 从积木库生成创意方向

# 生成3个创意方向(Markdown)
python spark.py

# 生成5个方向
python spark.py --count 5

# 指定话题过滤
python spark.py --topic "写作"

# 输出 JSON 格式(供 Agent 解析)
python spark.py --json

# 只用钩子生成
python spark.py --hook

# 固定种子(可复现)
python spark.py --seed 42

3. 生成完整文章

# 公众号长文
python write.py --topic "写作" --platform wechat --length 2000

# 小红书笔记
python write.py --topic "AI" --platform xhs --length 800

# 推文
python write.py --topic "自律" --platform twitter

# 输出 JSON 格式(包含 prompt)
python write.py --topic "个人品牌" --json

4. Agent 使用方式

Agent 读取脚本输出后,自行调用模型处理:

# 示例:Agent 读取 spark.py 的 JSON 输出
result = subprocess.run(["python", "spark.py", "--json"], capture_output=True)
ideas = json.loads(result.stdout)

# 然后用 Agent 的模型生成文章
for idea in ideas:
    article = await agent.generate(idea)

Dan Koe 方法论速查

APAG 框架

维度作用关键问题
Attention钩子,抓住注意力如何让人停下来?
Perspective描绘敌人,指出错误观念读者现在的错误想法是什么?
Advantage描绘英雄,提出正确视角如何让他们看到新的可能?
Gamify给出可操作步骤清晰、具体、下一步是什么?

POV > Niche

  • Niche 陷阱:选垂直领域 → AI 更快更好 → 被替代
  • POV 本质:你是 lens(透镜),不是垂直领域专家
  • 为什么不可复制:你的经历 + 目标 + 价值观的独特折射,AI 无法复制

积木类型使用指南

  • 钩子:用在标题、第一行,制造停顿
  • 悖论:放在开头或转折处,引发"这不对啊"的反应
  • 转化弧:规划文章整体故事线
  • 核心问题:戳中读者痛点引发共鸣
  • 金句:收尾或关键转折,制造记忆点
  • 结构:选择适合内容类型的框架
  • 观点:贯穿全文的独特视角,差异化核心

写作事业 6 步

  1. 选一个你无法停止谈论的话题
  2. 头脑风暴,提出独特观点
  3. 每周写 500-1000 字(短+长结合)
  4. 把长文拆解为每日短帖
  5. 学会让作品被分享(流量 > 写作)
  6. 用经验变现

短内容写作关键

  • 不是消费者,是研究者:看到好帖子试着用自己的观点重写
  • 结构训练:模仿句式 → 转化为自己的声音
  • 积木组合:同一金句/悖论可用于不同话题的写作

技术说明

  • 模型:使用 OpenClaw Agent 配置的模型(openrouter/minimax/minimax-m2.7
  • 无需额外 API Key:脚本只做数据处理,LLM 调用由 Agent 完成
  • JSON 输出:所有脚本支持 --json 参数,方便 Agent 解析

Comments

Loading comments...