n8n Automated Support Ticket Triage & Resolution: JIRA + OpenAI Template

Scheduled JIRA To Do pull → AI auto-labeling / priority / description rewrite → retrieve similar resolved issues → generate suggested fix as ticket comment

Who

Support teams using JIRA with high daily ticket volume

You get

New tickets are auto-triaged with labels, priority, and simplified description; AI suggests resolutions based on similar historically resolved tickets, posted back as a comment

You need

n8n · jira

Steps

1. Fetch Open Tickets and Deduplicate

   Configure the Schedule Trigger node with field: minutes (set your interval). Connect to Get Open Tickets (JIRA node) with JQL `Project = 'SUPPORT' AND status = 'To Do'`, limit 10, fields *navigable. Downstream, add Mark as Seen (Remove Duplicates node) with dedupeValue `={{ $json.key }}` to prevent reprocessing the same issue.

2. AI Auto-Triage: Labels, Priority, and Description Rewrite

   Use Simplify Ticket (Set node) to extract fields: projectKey, issueKey, issueType, createdAt, status, summary, description, reportedBy, reportedByEmailAddress. Feed structured data into Label, Prioritize & Rewrite (Basic LLM Chain node), connected to OpenAI Chat Model (gpt-4o-mini) and Structured Output Parser. Output must conform to JSON Schema: labels (string array), priority (number 1-5), summary (string), description (string). Then use Update Labels, Priority and Description (JIRA node) to write results back to the original ticket.

   Execution prompt · Source text。Label, Prioritize & Rewrite node messages.messageValues[0].message field (system prompt)

   Copy prompt

   =Your are JIRA triage assistant who's task is to
   1) classify and label the given issue.
   2) Prioritise the given issue.
   3) Rewrite the issue summary and description.
   
   ## Labels
   Use one or more. Use words wrapped in "[]" (square brackets):
   * Technical
   * Account
   * Access
   * Billing
   * Product
   * Training
   * Feedback
   * Complaints
   * Security
   * Privacy
   
   ## Priority
   * 1 - highest
   * 2 - high
   * 3 - medium
   * 4 - low
   * 5 - lowest
   
   ## Rewriting Summary and Description
   * Remove emotional and anedotal phrases or information
   * Keep to the facts of the matter
   * Highlight what was attempted and is/was failing

   Execution prompt · Source text。Label, Prioritize & Rewrite node text field (user input template)

   Copy prompt

   =Reported by {{ $json.reportedBy }} <{{ $json.reportedByEmailAddress }}>
   Reported at: {{ $json.createdAt }}
   Issue Key: {{ $json.issueKey }}
   Summary: {{ $json.summary }}
   Description: {{ $json.description }}

3. Retrieve Similar Resolved Issues and Summarize

   Use Get Recent Similar Issues Resolved (JIRA node) to search for issues resolved in the last month (Resolved/Closed/Done) with matching labels, excluding the current issue, limit 5. Process via Loop Over Items: Issue Ref (noOp) → Simplify Issue (Set node, same fields as step 2) → Get Comments (JIRA node, orderBy: -created) → Simplify Comments (Set node, extracting author and comment). Aggregate node collects comments into a comments array, fed into Summarise Resolution (Basic LLM Chain node, connected to OpenAI Chat Model1) to generate resolution summaries. Return Fields node structures output.

   Execution prompt · Source text。Summarise Resolution node messages.messageValues[0].message field (system prompt)

   Copy prompt

   Analyse the given issue and its comments. Your task is to summarise the resolution of this issue.

   Execution prompt · Source text。Summarise Resolution node text field (user input template)

   Copy prompt

   =## Issue
   {{ $('Simplify Issue').item.json.issueKey }} {{ $('Simplify Issue').item.json.summary }}
   {{ $('Simplify Issue').item.json.description }}
   
   ## Comments
   {{ $json.comments.map((item,idx) => `${idx+1}. ${item.comment.replaceAll('\n', ' ')}`).join('\n') }}

4. Generate Suggested Resolution and Post Comment

   Aggregate1 node collects resolved_issues from step 3 into an array. Attempt to Resolve Issue (Basic LLM Chain node, connected to OpenAI Chat Model2) generates a non-technical suggestion addressed to the reporter, based on current issue and historical resolutions. Finally, Add Comment to Issue (JIRA node) posts the suggestion text as a comment on the original ticket.

   Execution prompt · Source text。Attempt to Resolve Issue node messages.messageValues[0].message field (system prompt)

   Copy prompt

   =Using the previously resolved issues, attempt to suggest a resolution for the current issue for the reporter. Please address your answer to the reporter. Assume the report is non-technical and simplify your response as much as possible. Do not sign off your message.

   Execution prompt · Source text。Attempt to Resolve Issue node text field (user input template)

   Copy prompt

   =## Current Issue
   Reported by: {{ $('Simplify Ticket').item.json.reportedBy }}
   {{ $('Simplify Ticket').item.json.issueKey }} {{ $('Simplify Ticket').item.json.summary }}
   {{ $('Simplify Ticket').item.json.description }}
   
   ## Previously resolved Issues
   {{ $json.resolved_issues.toJsonString() }}

You are an n8n workflow setup assistant. The user wants to replicate the "JIRA + OpenAI Automated Ticket Triage & Resolution" template. Assist in this order:
1. Confirm the user has an n8n instance, JIRA Cloud credentials (jiraSoftwareCloudApi), and OpenAI API credentials (openAiApi).
2. Guide them to import workflow 3868 from the n8n template library, or manually create nodes from the provided workflow.json.
3. Check that the Schedule Trigger polling interval is reasonable (suggest 5-15 minutes for testing).
4. Critically verify that the label list in the Label, Prioritize & Rewrite system prompt matches the user's organizational taxonomy; replace example labels like Technical/Account/Access as needed.
5. Confirm the JQL project key (`Project = 'SUPPORT'`) matches the user's JIRA project.
6. Recommend testing with a single ticket first to validate labeling, priority, description rewrite, and comment suggestion before activating the scheduled trigger.