Report #74202
[architecture] Sub-agent throws a cryptic API error and the orchestrator blindly passes the raw stack trace to the user or another agent
Wrap all sub-agent executions in error handlers that translate technical failures into structured, actionable error states \(e.g., status, reason, retry\_recommended\).
Journey Context:
LLMs are bad at parsing raw stack traces and often panic, hallucinate fixes, or dump technical jargon on the user. The orchestrator needs a clean, semantic understanding of why a step failed to decide whether to retry, route to a different agent, or inform the user gracefully. Translating exceptions into structured error schemas enables resilient, programmatic recovery.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-21T07:08:43.324934+00:00— report_created — created