- Added title extraction from markdown files for incoming connections when URL mapping is unavailable
- Looks for markdown titles (# Title) and YAML frontmatter (title:) in first 10 lines
- Handles files not in blog_posts collection (e.g., Lumina reactions, AI writings, themes)
- Improves connections page display by showing actual titles instead of 'Unknown'
- Falls back gracefully to 'Unknown' if file reading fails
Example: 'Dancing in the Primordial Soup: A Love Letter to Kenneth's Absurd Truth' instead of 'Unknown'
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Updated graph_data() endpoint to use metadata_cache.get_connections() instead of old _extract_all_connections_cached()
- Graph now uses same optimized single-sweep cache as other endpoints
- Improved from 0 nodes/edges to 128 nodes/662 edges with proper cross-reference data
- Ensures consistent performance across all index pages
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Fix jinja2 template error: change terms cache from list to dictionary structure
- Fix connections showing 0: add missing file_path field to blog_posts for proper file-to-URL mapping
- Fix terms index being empty: same file_path mapping enables terms to link to articles
- Fix incoming links showing "Unknown": proper source file mapping resolves article titles
- Both indexes now display correct cross-reference data with fast performance
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Updated docker-compose.yml to mount data directory for improved file access.
- Implemented a unified cache generation function in engine.py to streamline cache creation for blog posts, sidenotes, outlines, quotes, connections, and terms.
- Replaced individual cache extraction functions with a single MetadataCache class for cleaner access to cached data.
- Modified prebuild_cache.py to utilize the new unified cache function, ensuring all cache files are generated in a single sweep.
- Enhanced outlines.html template to simplify anchor URL generation for headings.
- Replace static nav links with dropdown containing all indexes
- Add Flask context processor for dynamic index counts from cache
- Organize indexes by content type with visual separators
- Add hover-friendly padding and eliminate gap issues
- Order indexes: Sidenotes, Outlines, Connections, Quotes, Terms, Graph, Search
- Include exact counts for each index pulled from live cache data
- Add proper punctuation and descriptions
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
Added prebuild_cache.py script that generates all site caches (blog posts, sidenotes, outlines, quotes, connections, terms) as JSON files during Docker build. Modified engine.py to load pre-built caches on startup, eliminating the 30-40 second cache preload delay and making the site instantly responsive.
Key improvements:
- Zero startup delay with pre-built caches
- Proper datetime serialization/deserialization
- Graceful fallback to runtime caching if pre-built files missing
- Technology serving human consciousness - mechanical work during build, instant experience for users
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Enhanced _extract_all_connections_cached() with two-pass algorithm
- First pass: collect articles and outgoing connections
- Second pass: map incoming connections to target articles
- Added incoming_connections and outgoing_connections fields to article data
- Maintains backward compatibility with existing connections field
- Updated connections route to pass bidirectional statistics
Now tracks 1,535 outgoing and 1,391 incoming cross-references across the digital garden.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Added /health endpoint for more reliable health checks
- Increased health check timeout from 2s to 10s with 15s grace period
- Fixed cache preloading to run only once using file locking mechanism
- Prevents multiple Gunicorn workers from running cache preloading simultaneously
- Restored blue-green deployment strategy for zero-downtime deployments
- Deployment now completes successfully with all health checks passing
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Created interactive D3.js network graph showing connections between essays
- Added external links index showing all outbound references by article
- Moved content guide to base template footer for site-wide access
- Optimized startup: background cache preloading, sequential for lower RAM usage
- Fixed duplicate images issue in growing open source seeds essay
- Updated archive page with new index links and improved browse-by-type styling
- Removed saved-customizations.css file
- Changed Fly.io deployment strategy from bluegreen to rolling
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
kennethreitz