Network interruption recovery
AIgency classifies DNS, connection, and timeout failures separately from generic process crashes so they can be retried on their own terms.
AgentDock Pro
Network interruption recovery is part of AIgency, available only in AgentDock Pro.
A dropped connection mid-turn isn't the same kind of failure as a genuine task or code error, and AIgency treats it differently.
What gets classified as a network interruption
DNS resolution failures, connection reset or refused errors, timeouts, host/network-unreachable errors, and generic fetch/socket failures are recognized as network conditions rather than deterministic task failures.
Why the distinction matters
A network interruption is treated as recoverable the same way a capacity condition is: it doesn't necessarily need to burn through the task's retry budget the way a genuine execution failure would, since the underlying work was never actually wrong — the connection just dropped. Once connectivity returns, retrying the same task is usually the right response rather than treating it as a defect requiring a fix.