Runbook: Webhook Retry Backlog

Project-Agent operations docs/ops/runbooks/webhook-retry-backlog.md

Severity: SEV-2; escalate when dead letters accumulate for customer-critical integrations.
Page: support on-call.
Triage:
Inspect /v1/ops/webhook-deliveries and /v1/ops/queue-health.
Confirm whether failures are isolated to one endpoint or global.
Validate signature expectations and downstream status codes.
Mitigation:
Disable webhook fanout if a global issue is amplifying retries.
Replay dead letters only after the destination is verified healthy.
Notify affected customers if approvals or governance exports are delayed.