← Back to Channels and integrationsOpenClaw channels
Channel recovery patterns
When channel behavior breaks, the right recovery depends on whether the failure is transport, visibility, policy, or expectation mismatch.
Recovery ruleRecover the failing layer first. Do not rotate secrets just because a room-level policy mistake looked scary.
Recovery classes
The main channel failure classes
Transport brokeToken, session, homeserver, or linked-device state is actually unhealthy.
Visibility brokeDM works, but room or shared-space delivery does not.
Mention policy confused peopleThe bot is healthy, but shared-space replies still require an explicit trigger.
Correct recoveries
What to do first in each class
Transport brokeCheck token/session health and runtime access before touching policy layers.
Visibility brokeTest DM vs shared-space behavior separately and inspect room/channel permissions.
Mention policy confused peopleTest an explicit mention before claiming the bot is dead.
Open deeper reference links for this page