The original title is a blog-style headline about shipping early and bugs found. Let me rewrite it for a mobile feed.

Short summary

A founder shares two bugs found by strangers within a week of shipping PulseWatch early. The first: a second /start ping before a run closes leaves orphaned runs and false alerts — fixed by superseding the open run. The second: flaky jobs cause alert fatigue — fixed by counting state transitions in a rolling window and entering a distinct flapping state that fires one entry alert and one settle alert instead of spamming every transition.

• •Shipping early surfaced two real bugs from strangers within days
• •Duplicate /start pings now supersede open runs instead of orphaning them
• •Flapping detection suppresses per-transition alerts, sending only entry and settle notifications

Generated with AI, which can make mistakes.