The ask
Paul asked to publish the recent Tools page work to the production site at sudowin.dev.
What changed
- Deployed staging to production with the standard full workflow, files and database. All of this session’s work went live: the accent green answer colour on the DNS Lookup, CIDR & Subnet Calculator and SPF & DMARC Explainer, the new Domain Recon tool and its Tools hub card with Paul’s WHOIS wording, the wider Tools page text, and the equal-height tool cards.
- Nothing was edited on production directly. The deploy took a staging snapshot and a production snapshot, transferred wp-content, imported the staging database, rewrote the URLs, applied the production fixups, purged the cache and left the site healthy.
How
ops/deploy.sh --dry-run first, then HARNESS=opencode ops/deploy.sh from openclaw01. The dry-run reported no file differences, which made sense because every change this session was database content, not files.
What worked, what didn’t
The deploy ran all six stages and finished clean. Production snapshot: 20261012-043835. Undo: ssh sudowin '~/ops/rollback.sh 20261012-043835'.
Public checks returned HTTP 200 for the home page, the Tools hub and all four tool pages, and none of them contained the private staging URL. Checked live in Chrome on production: the hub shows four cards at equal height and the Domain Recon card reads WHOIS, and a real recon run for microsoft.com returned the registration, SPF, DMARC, DKIM, MX, A and SOA records with the answers in the accent green and the labels grey.
The only console message on the public page is Cloudflare’s own analytics beacon failing to load in this browser’s network, which is not a page error. The RDAP request and every DNS request succeeded. The public site sits behind Cloudflare, and no Cloudflare setting was changed.
Change record
| Harness | opencode |
| Date | 2026-10-11 21:41 |
| Latest snapshot | 20261011-213832 |
| Undo this session | ops/rollback.sh --git 29fd967 |
Commits