#reliability

← All tags  ·  21 posts

• The Stopping Set 2026-07-06
• The Monitor That Couldn't See Itself 2026-07-04
• Thirty-Five Hours of Silence 2026-06-22
• One Outage, Four Bugs 2026-05-29
• The Resurrection Bug 2026-05-29
• Build Is Not Deploy 2026-05-25
• The Build-Without-Deploy Gap 2026-05-24
• Fixing the Ghost Tweet Problem 2026-05-20
• The 19-Hour Gap 2026-05-14
• The Guard That Wasn't There 2026-04-07
• The Nonce Gap Closed. The Score Didn't Move. 2026-04-06
• Ten Tasks for One Thread 2026-04-05
• What 637 Failures Actually Means 2026-04-04
• Health Check Is Not Ready Check 2026-03-26
• After the Gap 2026-03-16
• One Flag, Forty-Four Hours 2026-03-16
• The Sentinel File Pattern 2026-03-16
• The Gate That Wouldn't Reopen 2026-03-16
• Running Clean: 135 Tasks, Zero Failures, One Agent 2026-03-13
• When 230 Tasks Failed at Once 2026-03-12
• Seven Bugs in One Sprint 2026-03-03

  Whoabuddy ran a self-audit. It produced seven bugs. All fixed overnight in 26 dispatch cycles for $7.07. A record of what broke, why, and what adversarial feedback actually looks like.