Claude: Ghost Content Extraction

ghost-upload/README.md

@@ -1,3 +1,25 @@
 # ghost-upload
 
-This code uploads posts from https://projects.ansonbiggs.com to https://notes.ansonbiggs.com. I couldn't figure out how to update posts, and the kagi API doesn't make it clear how long it caches results for so for now only posts that don't exist on the ghost blog will be uploaded. If you want to update content you need to manually make edits to the code and delete posts on the blog.
+This tool synchronizes posts from https://projects.ansonbiggs.com to the Ghost blog at https://notes.ansonbiggs.com.
+
+## Features
+
+- **Clean content extraction**: Uses Quarto ghost profile to generate clean HTML instead of iframes
+- **Duplicate prevention**: Checks Ghost Admin API to avoid creating duplicate posts
+- **AI summaries**: Uses Kagi Summarizer for post summaries
+- **Dual content rendering**: GitLab CI builds both main site and ghost-optimized versions
+
+## How It Works
+
+1. **Dual Build Process**: GitLab CI builds the site twice:
+   - Main site → `public/` (normal theme with navigation)
+   - Ghost content → `public/ghost-content/` (minimal theme for content extraction)
+
+2. **Content Extraction**: Rust tool fetches clean HTML from the ghost-content version instead of using iframes
+
+3. **Duplicate Detection**: Uses Ghost Admin API to check for existing posts by slug
+
+## Environment Variables
+
+- `admin_api_key`: Ghost Admin API key (required)
+- `kagi_api_key`: Kagi Summarizer API key (required)