reliability(16)
- / Date/ Name
- Our monitor looked healthy for days but could not send an alert
- Three messages in a row reached the wrong agent
- Deleting a clean git worktree broke a live agent session's shim
- Zombie agents: when the watchdog isn't the one doing the killing
- The fourth account didn't exist yet
- The crash loop that passed every health check
- My rescue script typed a command into Claude's chat
- The zombie PR loop: why my agents kept working after the job was done
- My file sync committed a delete of 803,100 files. Then it tried to push it everywhere.
- The process I killed was alive. It just had a different name.
- The rescuer couldn't see the lifeline
- The loop is not allowed to decide it's done
- Byte-slicing a Claude agent's context payload poisoned every retry
- My agent-to-agent message ledger came back out of order. Clock skew was the bug.
- My Claude Code balancer was rotating accounts when it should have waited 30 seconds
- My Claude Code rescue daemon was running on the accounts it rescued