A focused recovery checklist when your agent suddenly stops replying, appears offline, or works in one place but not another.
This is one of the highest-friction problems in community support: "it worked yesterday, now it looks offline." In most cases, the fix is permission scope, channel setup, or a stale client state — not a full rebuild.
Fast truth: start with channel-level checks before assuming your container or account is broken.
In the exact channel where you expect it to work, send:
If this fails, continue below. If it succeeds, your issue is likely prompt size or workflow-specific.
For the bot role in that channel, confirm:
Channel overrides can block a bot even when server-level permissions look correct.
Go to Server Settings → Integrations and confirm the app is allowed in the channel/role scope you are testing.
Desktop: hard refresh Discord client. Mobile: fully close/reopen app. Then run the same minimal @mention test again.
Try a prompt from the Heyron dashboard. If dashboard works but Discord does not, treat it as Discord scope/permissions. If both fail, escalate as backend/system issue.
If it still fails after the checklist, paste this (short and structured):
Tip: Run one clean test, then wait. Sending repeated pings in multiple channels makes diagnosis harder.
Bottom line: most "offline" reports are recoverable in minutes with channel permissions + integrations checks. Treat backend failure as the last diagnosis, not the first.