Summary

Sam asked to use Daily Repo Health Check - 2026-05-10 as the source of truth and clean up the highest-signal failures in /Users/samaguiar/Documents/Projects with reversible fixes only.

What shipped

• Archived duplicate workflow file from .github/workflows/monthly-backlink-check 3.yml to admin/_archive/workflows/2026-05-10/monthly-backlink-check 3.yml. The active workflow folder now has only .github/workflows/monthly-backlink-check.yml.
• Moved the broken empty checkout from Repos/sail-googleads/google_ads_mcp to admin/_archive/google_ads_mcp-broken-checkout-2026-05-10/ and replaced it with a relative symlink to ../../google_ads_mcp.
• Patched routines-build/routines/github/schema_validator/run.py so image and other non-HTML sitemap URLs are skipped, first baseline runs do not count every existing issue as newly broken, and Notion logging failure is non-fatal after the report is written.

Root cause notes

• Heartbeat Check failed because it correctly detected .github/workflows/weekly-non-indexed.yml as stale. Latest heartbeat log showed last success never, age about 443h, threshold 192h. The underlying weekly non-indexed workflow has only failure or cancelled runs. The latest scheduled run on 2026-05-04 was cancelled at the workflow 30-minute timeout.
• Monthly Schema Validator failed because the validator completed and returned critical. Live reproduction showed 1,385 sitemap entries, 779 were media or other non-HTML URLs that the old validator tried to treat as pages. On the 606 real HTML URLs, there are still 606 schema validation warnings, especially LegalService.telephone, Person.name, and LegalService fields on some pages.
• The schema routine also had no persistent prior baseline in git because routines-build/state/*.json and date-based outputs are ignored, so an empty prior state made all existing issues look newly broken.

Verification

• find .github/workflows -maxdepth 1 -type f -name '*backlink*' returned only .github/workflows/monthly-backlink-check.yml.
• readlink Repos/sail-googleads/google_ads_mcp returned ../../google_ads_mcp.
• PYTHONPATH=Repos/sail-googleads/google_ads_mcp ... import ads_mcp.server_stdio succeeded and confirmed all six custom Google Ads write tools are present.
• routines-build/.venv/bin/python -m compileall -q routines-build/routines/github/schema_validator/run.py passed.
• Running the schema validator with Notion disabled wrote routines-build/outputs/2026-05-10/schema-validator.md and exited 0 with severity warning, not critical.

Remaining dirty-tree risk, separate from this fix

The project root was already dirty and remains risky to commit as one bundle. Current tracked dirty set includes 23 modified tracked files; only routines-build/routines/github/schema_validator/run.py was intentionally changed by this session. Large untracked areas remain, including google_ads_mcp/, many Skills folders, Repos/sail-knowledge/, Repos/sail-spanish-lp/, and admin/_archive/.

Reason for ending

Requested reversible cleanup and verification are complete. Follow-up is not blocked, but there are approval-worthy cleanup choices for the remaining dirty tree and durable routine behavior.