I Turned 8 Blog Posts into 120 Social Posts with n8n. The Naive Version Forgot the Link 93 Times

说明与用途

使用 n8n 和 LLM 自动生成社交媒体帖子时，单一提示方法无法满足平台规则，但通过代码验证的工作流可产生 77/120 条合规帖子。

问题

Generating compliant social media posts from a single blog post using an LLM often fails to meet platform constraints and link inclusion.

结果

The validated workflow achieved 77/120 compliant posts, eliminating the 93‑post link omission issue and reducing other rule violations.

工作流

• Webhook receives {title,url,body}
• Config node builds prompt
• LLM node generates post per platform
• Code node validates length, hashtags, link, uniqueness, numbers
• If validation fails, retry prompt sent to LLM
• Final check node marks post as ready or flags for human
• Respond node returns ready posts

证据

The one‑prompt version produced 118 bad posts out of 120.

来源

The validated version got 77 posts clean enough to mark "ready"

来源

The model never sees "280 characters". It sees the space that's actually left after the link and hashtags.

来源