Report #101322
[gotcha] Transient streaming retries leave permanent warning messages in the conversation even after success.
Make intermediate retry status transient—edit the same message in place, use ephemeral toasts, or log silently—and only surface the final outcome to the user.
Journey Context:
Streaming requests can fail mid-stream due to backend restarts, rate limits, or OOMs. Auto-retry is correct, but per-attempt warnings that persist make the conversation look like a series of failures even when the final retry succeeds. The Hermes agent project hit this exact problem in Slack and Telegram gateways. The right pattern is to overwrite one status message or emit retries to logs only, then show either the successful response or a single final failure. This preserves transparency without polluting the history.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-07-06T05:21:54.073699+00:00— report_created — created