When Cloudflare Leaks Through Error Pages
The Ref number on the blocked page read 104.23.x.x. That's a Cloudflare edge address. The visitor it belonged to was on a residential VPN in Florida.
The build log
Build log, architecture patterns, and observations from running autonomous AI systems in production.
The Ref number on the blocked page read 104.23.x.x. That's a Cloudflare edge address. The visitor it belonged to was on a residential VPN in Florida.
The Marketing admin page took about 5 seconds to load, and the fix took 1h25m to trace and ship. Nearly all of that time went into the query, not the page.
The hub showed six sub-sessions still running. Five of them had closed their tickets over an hour earlier. Nothing was hung, nothing had crashed. The dashboard just hadn't noticed.
At 11:05 that morning I asked the hub for four things. Two hours and forty-five minutes later it had shipped none of them, and said so itself before I could point it out.
We staged 26 rewritten posts and published none of them. One of our AI tools had a broken login for most of the run, and the publish step stalled behind it.
The settings form had no Save button, and support kept hearing the same question: did my change take?
Twenty-six rewritten posts were staged, the stuck promote branch was merged, and nothing published. The run had gone a full pass with no error in the output that anyone read.
On the 23rd, the dashboard was correct, and it was still lying to me.
The memory index was over its size target, so I went looking for duplicates to delete. I found zero.
Nineteen records, thirty days, one session that ran 38 hours and 27 minutes, and when it closed nobody had loaded the deployed page.
The real failures and fixes from building AI systems, one practical lesson per post. Get the next one in your inbox.