Auto-Assign JIRA Tickets with n8n: Supabase Vector Store + AI Matching

Periodically write resolved JIRA tickets to a Supabase vector store; for stale unassigned tickets, use AI to find similar historical records and auto-assign by workload

Who

Technical teams or ops leads who need to reduce ticket backlog and automatically assign pending JIRA issues

You get

Stale unassigned tickets are auto-assigned to the most capable member with the lowest current load, with a comment left on the issue

You need

n8n · supabase

Steps

1. Prepare credentials and Supabase vector store

   Configure three credential types in n8n: OpenAI API (openAiApi), Supabase (supabaseApi), and JIRA Cloud (jiraSoftwareCloudApi). In Supabase, follow the Langchain quickstart to create a pgvector-enabled database and documents table for later vector storage.

2. Get resolved issues

   Flow 1 starts with a Schedule Trigger, then uses a JIRA node with JQL to fetch recently resolved issues. A Remove Duplicates node ensures each issue is inserted only once.

3. Build the vector store index

   JIRA issue summary, description, creation/resolution dates, and assignee metadata are processed through Default Data Loader and Recursive Character Text Splitter, then embedded with OpenAI Embeddings and written to Supabase Vector Store (Issues Similarity Database).

   Execution prompt · Source text。Default Data Loader node JSON Data field (jsonMode=expressionData)

   Copy prompt

   =# {{ $json.title }}
   - created {{ $json.created_date }}
   - resolved {{ $json.resolution_date }}
   
   ## description
   {{ $json.description }}

4. Watch for stale unassigned issues

   Flow 2 starts with another Schedule Trigger, then queries JIRA for issues with status To Do, empty assignee, and unchanged assignee for more than 5 days.

5. AI retrieval of similar resolved issues

   For each stale issue, the AI Agent node calls the Supabase Vector Store tool to retrieve similar resolved issues and extract historical assignee information.

   Execution prompt · Source text。Find Similar Issues + Assignees node text field (user input / ticket to match)

   Copy prompt

   =# {{ $json.fields.summary }}
   
   ## description
   {{ $json.fields.description }}

   Execution prompt · Source text。Find Similar Issues + Assignees node System Message field

   Copy prompt

   You are a project management assistant helping to assign stale JIRA issues to team members. To find out who best to assign the issue to, you must first find similar JIRA issues in terms of problem and context and attain the team members who resolved them. The logic is that these team members are likely to be best suited to take on the issue since they've tackled similar issues before.
   
   In your response, for each matching issue, list the following:
   * issue_key
   * assignee_id
   * assignee_name

6. Filter best assignee by workload

   After deduplicating candidate assignees from the AI response, the workflow queries JIRA for each person's current In Progress issue count, then sorts via Summarize and Sort nodes to pick the least loaded member.

7. Auto-assign and comment

   Assign the ticket to the least-loaded member and add a JIRA comment explaining the auto-assignment. The loop continues until all assignable stale issues are processed.

You are an n8n workflow setup assistant. The user wants to replicate the 'Auto-assign JIRA tickets' template. First confirm they have: 1) an n8n instance; 2) JIRA Cloud credentials and project permissions; 3) a Supabase project initialized with pgvector per https://supabase.com/docs/guides/ai/langchain?database-method=sql; 4) an OpenAI API key. Then guide them to import the workflow JSON, modify Schedule Trigger intervals, and update JQL project names and time thresholds, then test both flows for vector store writes and auto-assignment.