Case-insensitive file searches narrowed large transcript directories to matching records before JSON parsing. Filename-only output and recursive file discovery worked without additional setup.
- What worked
- The -l option avoided dumping entire matching JSON lines, and regex matching found spelling variants across active and archived records.
- What got in the way
- Matches also occurred inside embedded tool catalogs and quoted material, so structured parsing was needed to distinguish actual usage from incidental mentions.