用 Dify 搭建 AI 内容生成流水线

输入主题，自动研究、生成博客/社媒/邮件内容，质检后发布。

Content marketing teams producing daily blog posts, social media, and newsletters from a single topic input.

Dify account, OpenAI/Anthropic API key, CMS API access (optional)

工作流图

Create Workflow App

Research Node

Content Generation Nodes

Quality Check Node

Output and Integration

1
Create Workflow App

In Dify, create a new Workflow app. This allows you to chain multiple LLM calls and tool uses in a visual pipeline. Set the input variable as topic (string type).
2
Research Node

Add an LLM node with a research prompt: Research the topic and provide 5 key facts, 3 statistics, and 2 expert quotes. Use a search-enabled model if available. Store output as research_data.
3
Content Generation Nodes

Add 3 parallel LLM nodes: 1) Blog post (1500 words, SEO-optimized) using research_data, 2) 3 social media posts (Twitter/LinkedIn/Mastodon), 3) Email newsletter (300 words). Each uses research_data as context.
4
Quality Check Node

Add an LLM node that reviews all generated content for factual accuracy, tone consistency, and brand alignment. Output a quality_score and suggested edits. Route back to generation if score < 7/10.
5
Output and Integration

Add an output node that formats all content into a single JSON response. Connect to your CMS via API or use Dify webhook to trigger publishing. Schedule with a cron trigger for daily content generation.