使用 Qwen 3 构建多代理图书作家

Write 20k word book from 3-5 words.

从 3-5 个词撰写 20k 字书籍。

Today, we are building an Agentic workflow that writes a 20k-word book from a 3-5 word book title.

今天,我们正在构建一个 Agentic workflow,它能从一个 3-5 个词的书名生成一本 20k 字的书籍。

Tech stack:

技术栈:

  • Firecrawl for web scraping.
  • Firecrawl 用于网页抓取。
  • CrewAI for orchestration.
  • CrewAI 用于编排。
  • Ollama to serve Qwen 3 locally.
  • Ollama 用于本地服务 Qwen 3。
  • LightningAI for development and hosting.
  • LightningAI 用于开发和托管。

Here's our workflow:

这是我们的工作流程:

  • Using Firecrawl, Outline Crew scrapes data related to the book title and decides the chapter count and titles.
  • 使用 Firecrawl,Outline Crew 抓取与书籍标题相关的数据,并决定章节数量和标题。
  • Multiple writer Crews work in parallel to write one chapter each.
  • 多个作家 Crews 并行工作,每个写一章。
  • Combine all chapters to get the book.
  • 将所有章节组合起来得到书籍。

The code is available in this Studio: Book writer flow. You can run it without any installations by reproducing our environment below:

代码在这个 Studio 中可用:Book writer flow。您可以无需任何安装,通过复制下面的环境来运行它:

Let’s implement this!

让我们来实现这个!


Scraping tool - SERP API

抓取工具 - SERP API

Books demand research. Thus, we'll use Firecrawl’s SERP API to scrape data.

书籍创作需要研究。因此,我们将使用 Firecrawl’s SERP API 来抓取数据。

Tool usage:

工具使用:

  • Outline Crew → to research the book title and prepare an outline.
  • Outline Crew → 研究书籍标题并准备大纲。
  • Writer Crew → to research the chapter title and write it.
  • Writer Crew → 研究章节标题并撰写它。

Set up Qwen 3 locally

本地设置 Qwen 3

We'll serve Qwen 3 locally through Ollama.

我们将通过 Ollama 在本地提供 Qwen 3。

To do this:

要做到这一点:

  • First, we download it locally.
  • 首先,我们将其下载到本地。
  • Next, we define it with the CrewAI's LLM class.
  • 接下来,我们使用 CrewAI 的 LLM 类定义它。

Outline Crew

Outline Crew

This Crew has two Agents:

这个 Crew 有两个 Agents:

  • Research Agent → Uses the Firecrawl scraping tool to scrape data related to the book's title and prepare insights.
  • Research Agent → 使用 Firecrawl 抓取工具抓取与书籍标题相关的数据并准备见解。
  • Outline Agent → Uses the insights to output total chapters and titles as Pydantic output.
  • Outline Agent → 使用这些洞见输出总章节数和标题,作为 Pydantic 输出。

Writer Crew

Writer Crew

This Crew has two Agents:

这个 Crew 有两个 Agents:

  • Research ...
开通本站会员,查看完整译文。

- 위키
Copyright © 2011-2026 iteam. Current version is 2.155.1. UTC+08:00, 2026-04-03 07:22
浙ICP备14020137号-1 $방문자$