Tickets convert a conversation request into durable support work. They carry a title, description, status, priority, origin, assignee, tags, notes, resolution metadata, and links back to the conversation/contact when available.
AI actions
Internal AI routes use AI_TOOL_SECRET, while operator routes use JWT authentication, organization resolution, and agent role checks. The console exposes ticket lists, detail, assignment, status updates, notes, and replies.
Ticket lifecycle
Priorities are low, medium, high, or urgent. Origins include widget, email, WhatsApp, Telegram, operator roles, AI, and API.
Idempotency and traceability
Ticket storage includes an idempotency-key index in metadata, which AI and integrations should use when retries could create duplicate work. Notes record author and author type. Agent-run steps should link tool execution to the originating conversation and message.
Models should not claim that a ticket was created, updated, or closed until the tool returns success. User-facing ticket identifiers must come from the gateway response.