Day 45 — Value Added, Not Volume | AI Education Project

A file-count of today's session would undersell it — a handful of edits across a handful of files. But a couple of those edits fixed navigation that was quietly broken for anyone who actually clicked through it, and a couple more closed items that had been sitting on the parking lot since Day 42. Small footprint, real debt paid down. That's the measure that matters today.

Verification, Not Just Trust

Re-checked all three of Day 44's self-reported cleanup items against the fresh tar (old commented banner, four superseded sequence-index drafts, two Bluefish backups) — all three confirmed genuinely clean, not just claimed clean

Real Bug Caught — Broken Diary Navigation

Two diary pages in building-in-public/ (Day 42 and the reconstructed Day 43) had four links each pointing one directory level too shallow — ../AI-NEW.html and index.html instead of ../../AI-NEW.html and ../index.html
Both top and bottom nav in both files fixed and confirmed

Tooling Hardened for the Future

Find-file-NO-breadcrumb.sh — scoped to *.html only, so non-HTML files stop getting flagged every run (closed a Day 44 follow-up item)
check_tree.sh — added -s/--silent to suppress the per-file source listing; added an automatic skip for any href containing a literal backslash (real hrefs never have one — this is always a code/regex sample false positive, caught live today in sed-examples-claude-dg.html); added an extensible IGNORE_HREF_PATTERNS array for future one-off false positives, matching the existing SKIP_NAMES/SKIP_EXTS convention

Parking Lot — Both Standing Items Closed

AD directory (carried since Day 42) — decided: removed from the site index, directory moved to saved/
Master index-of-indexes (carried since Day 42) — turned out to already exist as EXPANDED-index.html, just not fully wired up or discoverable. Closed the gaps: gave SED-AWK the index.html symlink every other section already had, updated its card to match, added a new "Already know what you need? Browse all sections directly." card to the root index.html, and fixed the page's breadcrumb depth bug (../index.html → ../../index.html)
Along the way: confirmed the three-tier structure was already sound — sequence-index.html (curated 9-stop path) → EXPANDED-index.html (comprehensive, all 22 sections) → Learn_by-catagory.html (topic browse) — each solving a different problem, not redundant with each other

Housekeeping

LearnLinux-Content-Worklist.html (a checkbox-tracker working doc) moved out of the site to SAVE, same exempt category as building-in-public/

Close-out: vnu clean (aside from known CSS-checker-lag false positives on @property/@layer/@container — real, modern, valid CSS). check_tree -a fully clean: 0 orphans, 0 broken links, 38 symlinks, 0 .html~ backups.

Files touched: not many. Debt paid down: real. That's a ++ day.