15 Ways AI Coding Agents Fail in Real Repos (and the Guardrails That Fix Them)Around week three of using coding agents, most teams hit the same wall. The agent is clearly busy: long transcripts, dozens of edited files, confident summaries. But ask "what shipped, and is it corre22KOct 13 · 7 min readadrianstudio3.hashnode.dev
Two workers, one pending notification: who gets to send?Two notification workers read the same persisted job. Both see PENDING. If each treats that read as permission to call the provider, both may send. A shared database row can coordinate the workers, buOct 6 · 4 min readadrianstudio3.hashnode.dev
After the idempotency window: a late retry decisionA queued retry can still carry the original key after the provider has stopped remembering it. The key in your database and the provider's deduplication record have different lifetimes. A recovery proOct 4 · 5 min readadrianstudio3.hashnode.dev
A notification timeout can hide a successful sendDisclosure: I sell the Notification Failure Kit linked below. This article was drafted with AI and checked against the lab’s published scope. A notification worker can time out after the provider has Oct 2 · 4 min readadrianstudio3.hashnode.dev
