Friday morning started with a bit of a visibility crisis. I looked at the social share cards for the various ops sites and realized most of them were non-existent. It’s one of those things that doesn’t show up in error logs but kills credibility the second someone drops a link in a Slack channel. I spent the first half of the day playing whack-a-mole with Open Graph tags across the ecosystem.
The pattern was pretty consistent. Most of the subdomains like harmony, email, customer, stable, and brigade just had zero OG tags. It was a quick fix but a necessary one. I added the social share cards to each of those repos so they would actually render a preview image when shared. For ops-center, the issue was slightly different. The OG image was absolute but the card itself wasn’t sized correctly. I fixed the real 1200x630 card there so the preview looks clean instead of stretched.
Then there was accounting-ops. That one was actually broken. The og:image was pointing to http://localhost:3000. Someone had probably copied a local config and forgot to change it before pushing. I repaired the dead social card and cleaned up the references in the layout and page files. Now when you share the accounting link, you get a proper image instead of a broken link icon. It’s a small thing but it matters.
I added the social share cards to each of those repos so they would actually render a preview image when shared.
After the social card sweep, I shifted gears to the wealth-ops repo. There was some serious plumbing work happening under the hood with the execution layer. I added test coverage for the kill-switch and credential-resolution guardrails. These are critical safety nets. The approve-path kill-switch needs to be bulletproof because it stops bad executions in their tracks. I wrote tests to make sure it actually triggers when it should.
The credential resolution part was trickier. We are in the middle of a migration, and legacy hand-wired plaintext credentials need to keep working until everything is fully switched over. If I broke that, people would lose access to their systems during the transition. I fixed the credentials service to handle both the old plaintext format and the new secure resolution path. It’s a bit of a hybrid state but it keeps things running smoothly for now. I added tests to verify that the legacy credentials still resolve correctly so I know I didn’t break backward compatibility.
It was a solid Friday. I got the visibility sorted across the board and tightened up the safety rails in wealth-ops. No major new features but a lot of polish and protection. The sites look better when shared and the execution layer is more resilient. That’s a good way to end the week.
Real product captures — click any to enlarge.