store: CLAUDE.md optimization: 329 → 104 lines with table-based loading rules

graph/solutions/claudemd-optimization-329-104-lines-with-table-based-loading-f3feda.md

@@ -0,0 +1,27 @@
+---
+id: f3fedac1-e6f9-4339-b8ee-619e2482f9e2
+type: solution
+title: "CLAUDE.md optimization: 329 → 104 lines with table-based loading rules"
+tags: [claude-config, optimization, documentation, claude.md]
+importance: 0.7
+confidence: 0.8
+created: "2026-02-13T21:14:57.451590+00:00"
+updated: "2026-02-13T21:14:57.451590+00:00"
+---
+
+Optimized project CLAUDE.md from 329 lines to 104 lines (68% reduction) by:
+
+1. Compressing repetitive loading rules into keyword table format
+2. Removing 72 redundant 'Note:' lines (info already in CONTEXT.md files being loaded)
+3. Stating loading convention once instead of repeating for each of 13 technologies
+4. Preserving all functionality: special cases (Pi-hole, n8n, Ko-fi, Jellyfin), maintenance protocol, warning triggers
+
+Key insight: The loading rules were verbose because each tech block repeated the same pattern:
+- Load CONTEXT.md (technology overview)
+- Load troubleshooting.md (error handling)
+- If in scripts/, load scripts/CONTEXT.md
+- Note: [details already in the files being loaded]
+
+Solution: State convention once at top, then use compact keyword→directory table. All inline notes removed since that info is in the actual CONTEXT.md files.
+
+Result: Much easier to scan, maintain, and add new technologies. No loss of functionality.