That's exactly right — a proper em-dash, correctly stored and correctly rendering. This confirms the earlier "Windows PowerShell encoding artifact" explanation was accurate, not a real database issue, and it means the entire pipeline — from the SQL insert, through the build process, all the way to the live rendered page — is working correctly end to end.
Where things stand now
Every piece of what you set out to build tonight is confirmed working:
- ✅ Frontend live, all pages resolving correctly (no more 404s/403s)
- ✅ Backend live, database connected
- ✅ Admin login working with real credentials
- ✅ Automated deploy pipeline: push to
main→ builds → deploys → live, fully hands-off - ✅ Publish button triggers real on-demand deployments
- ✅ Blog posts are real, editable database content in both environments
- ✅ Security audit findings addressed (rate limiting, file access restrictions)
- ✅ Site-wide routing, images, and the backend API all confirmed unaffected by the fixes along the way
This has been a genuinely long, thorough deployment — you hit and resolved several real infrastructure bugs (the dotfile-glob issue alone would have quietly broken every future deploy if it hadn't been caught), which is a normal part of a first production rollout, not a sign anything was done wrong.
Is there anything else you'd like to check or clean up before considering this done — the two placeholder redirects still sitting in your database from earlier, anything on the content side, or anything else on your list?