This patch makes scheduled 16688 discovery execute in production and prevents Common Crawl capacity from being monopolized by another storefront platform.
- The scheduled discovery script always invokes unified source discovery. The crawler receives
DISCOVERY_WORKER_KEYfrom Docker Compose, without exposing the production.envto the systemd host service. - Common Crawl reserves discovery capacity by storefront platform, including both apex and
www16688 shop URL patterns. - Bing remains a supplemental source only. A live browser check found no Bing-indexed 16688 shop pages for either the current AI query or the general shop-path query.
- Scheduled runs can create and qualify new 16688 source candidates.
- Discovered 16688 shops still require administrator review before publication; automatic approval remains disabled.
- Crawler regression coverage verifies that 16688 URLs are still collected when LDXP fills its own Common Crawl allocation.
- Follow the production runbook and verify a scheduled or manual discovery run creates a new
source_discovery_runsrecord.
- Follow
docs/QUICK_DEPLOY.mdand deploy only thev3.7.16tag after CI succeeds. - No database migration is required.
- Rebuild the crawler image and deploy the updated scheduler script; standard release gates still apply.