Tickets
This document explains how to create a new ticket in the DuploCloud AI Suite HelpDesk, configure the agent, scope, and skills, and interact with the AI agent on the resulting ticket page.
Prerequisites
Access to the DuploCloud AI Suite
An authenticated user account
At least one agent, scope, and skill configured in the system
Step 1 — Navigate to Add Ticket
Go to HelpDesk in the left-hand navigation and click Add Ticket, or navigate directly to the add ticket URL. The Create a new ticket form appears with fields for Select Agent, Select Scopes, and Advanced Options.
Step 2 — Select an Agent
Click the Select Agent button. A dropdown list appears showing all available AI agents. The option to select is highlighted below.
Click generic-agent to select it.
The agent determines which AI model handles your ticket. Different agents may have different specializations, tools, and capabilities.
Step 3 — Select a Scope
Click the Select Scopes button. A dropdown appears showing all scopes your account has access to. The option to select is highlighted below.
Click full-admin-token to select it.
A scope defines the set of credentials and permissions the agent will use when executing actions on your behalf (e.g., querying AWS, accessing Kubernetes clusters, calling APIs).
Step 4 — Open Advanced Options
Click Advanced Options to expand the advanced configuration section, as highlighted below.
This reveals two additional settings: Personas and Non Interactive.
Step 5 — Select All Personas
In the Personas section check all personas you want the agent to use. The personas selection is highlighted below.
Personas control which specialized knowledge modules the agent can invoke. For example, the devops persona enables CI/CD and Kubernetes guidance, while GCP-troubleshooting enables GCP-specific diagnostics.
Step 6 — Non Interactive Option (Optional)
The Non Interactive option is highlighted below. Leave it unchecked for a standard interactive session.
Note: Enabling Non Interactive causes the agent to run autonomously end-to-end without pausing for user input. Useful for fully automated workflows — leave unchecked for normal conversational use.
Step 7 — Enter Your Question
Click in the text area and type your question or task description. The textarea is highlighted below.
Step 8 — Submit: Create Ticket
Click the Create Ticket button, highlighted below, to submit the form.
The system processes the request and redirects you to the ticket detail page.
Ticket Page — Feature Overview
After creating a ticket you land on the ticket detail page. The page has an Activity Feed on the left for the conversation with the agent, and a Ticket Details sidebar on the right showing metadata.
Ticket Details
The Ticket Details heading (highlighted) marks the start of the sidebar that summarises all ticket metadata.
Status
The STATUS field (highlighted) shows the current state of the ticket — Open, In Progress, Resolved, or Closed.
Total Cost
The TOTAL COST field (highlighted) shows the cumulative token/compute cost consumed by the AI agent for this ticket.
Scopes
The SCOPES field (highlighted) lists the credential scope(s) attached to this ticket (e.g., full-admin-token). The agent uses these credentials when executing actions on your behalf.
Personas
The PERSONAS field (highlighted) lists the persona modules active for this ticket (e.g., devops, GCP-troubleshooting, GCP-IAC). These control which specialized knowledge the agent can draw upon.
Priority
The PRIORITY field (highlighted) indicates the urgency level of the ticket — Low, Medium, High, or Critical.
Ticket ID
The TICKET ID field (highlighted) is the unique identifier for the ticket in the format <WORKSPACE>-<NUMBER> (e.g., AIFULLINTERNAL-45). Use this ID to reference the ticket in logs, reports, or support conversations.
Ticket Secret
The Ticket Secret section (highlighted) displays a secret token associated with this ticket. This token can be used by external systems or webhooks to authenticate callbacks and API calls scoped to this specific ticket.
Command Permissions
The Command Permissions section (highlighted) controls what system commands the agent is allowed to execute during this ticket.
Clicking it opens the Command Execution Permission Settings modal shown below.
Inside the modal you can:
Approve All — allow all commands the agent wants to run
Reject All — deny all command execution
Add pattern for auto-approval — any command matching the pattern is automatically approved
Add pattern for auto-rejection — any command matching the pattern is always blocked
Context Files
The Context Files section (highlighted) allows you to attach files such as Kubernetes manifests, CI configs, or logs. The agent reads these as additional context when answering your question or executing tasks.
Reports
The Reports section (highlighted) shows any structured reports or outputs generated by the agent during this ticket, such as audit summaries, cost breakdowns, or diagnostic results.
Feedback List
The Feedback List section (highlighted) lets you rate and comment on the agent's responses. Feedback is logged and visible to administrators for quality review and model improvement.
Activity Feed
The Activity Feed on the left is where the conversation with the agent takes place. Your original question appears at the top, and the agent streams its response below it in real time.
Live File Diffs
When the agent creates, edits, or overwrites a file, the change appears in the Activity Feed as a diff instead of only a closing summary. The header shows whether the file was created, edited, or overwritten, along with +N/-N line-count badges — expand it to see the full line-by-line diff, with additions and deletions color-coded like a code review.
The diff appears once the file operation finishes — after the "Running tool" status, alongside the agent's usual explanation of what it changed and why — rather than streaming line-by-line as it's written.
Long Conversations and Context Compaction
Long-running conversations are automatically compacted as they approach the underlying model's context window limit. The agent summarizes the earlier turns of the conversation and continues from that summary, rather than losing history or failing outright once the limit is reached.
While compaction is running, two things change in the ticket view:
The Activity Feed shows a status line — "Compacting conversation history to fit the context window…" — followed shortly by "Compaction complete." Click the completed line to expand it and see an excerpt of the summary the agent generated.
The chat input's placeholder changes to "Compacting conversation history…", and the Stop button is disabled with the tooltip "Compacting — cannot stop."
Compaction is fully automatic — there is no setting to configure or threshold to tune. It is triggered by the underlying model's context window, not by a DuploCloud-side limit.
Playwright Test
Last updated
Was this helpful?