Updated the needs-followup classifier in categorize.py to better bucket SEO, Web Design, Google Ads, and Operations work. The targeted (general) bucket total dropped from 198 to 34 after validation against the live KB.
The classifier edits were implemented, syntax-checked, validated against the live backlog, and the dashboard was regenerated. No additional archive action was taken in this session.
The clearest remaining taxonomy drift is now in Litify and Admin/General. A future tuning pass there should be straightforward. The next scheduled triage run can safely pick up the 3 newly stale routine-log candidates surfaced by today’s dry run.
What shipped: classifier updates, validation run, refreshed dashboard, local comparison note, and QA reflection. What did not ship: any archive apply pass, any Litify/Admin retune, or changes to the archive thresholds.