Auto-classify GitHub issues and create Linear tasks

Listen for new GitHub issues → AI classification (type/priority/labels) → Create Linear task and post GitHub comment

Who

Teams using n8n that need automated GitHub issue triage synced to Linear

You get

New issues trigger automatic AI classification, Linear task creation, and GitHub comment feedback

You need

n8n · github · linear

Steps

1. Configure trigger and filter

   In n8n, set up the Github Trigger node: select the repository and subscribe to issues and issue_comment events. Add an If node with the condition $json.body.action equals 'opened' to process only newly opened issues. Add an Edit Fields (Set) node to extract and map fields: title = $json.body.issue.title, description = $json.body.issue.body, issue_url = $json.body.issue.url, author = $json.body.issue.user.login.

2. AI classification and formatting

   Add an Information Extractor node, connected to an OpenAI Chat Model sub-node (model options: gpt-4.1-mini / gpt-4.1 / gpt-4). Paste the classification prompt into the Information Extractor's text field. Add a Code in JavaScript node to parse and normalize the AI's JSON output, generating priorityLabel, typeLabel, and labels. Add a Merge node (mode: combineByPosition) to combine original issue data with the formatted AI output.

   Execution prompt · Source text。Information Extractor node text parameter field

   Copy prompt

   =You are an AI issue triage assistant.
   
   STRICT RULES:
   - Output ONLY valid JSON
   - No explanation
   - No markdown
   - No extra text
   
   Classify the issue:
   
   type must be one of:
   Bug | Feature | Question
   
   priority must be a NUMBER:
   1 = Urgent
   2 = High
   3 = Medium
   4 = Low
   
   Also generate 2–4 short labels.
   
   Return format:
   
   {
     "type": "Bug",
     "priority": 1,
     "labels": ["crash", "login"]
   }
   
   Issue Title:
   {{ $json.title }}
   
   Issue Description:
   {{ $json.description }}

   Execution prompt · Source text。Code in JavaScript node jsCode field

   Copy prompt

   let output = $json.output;
   
   // 1. Parse if AI returned string
   if (typeof output === "string") {
     try {
       output = JSON.parse(output);
     } catch (e) {
       throw new Error("Invalid JSON from AI");
     }
   }
   
   // 2. Normalize values
   const priorityNum = Number(output.priority);
   
   // 3. Fallback safety
   const safePriority = [1,2,3,4].includes(priorityNum) ? priorityNum : 3;
   
   // 4. Priority label mapping
   const priorityMap = {
     1: "Urgent",
     2: "High",
     3: "Medium",
     4: "Low"
   };
   
   
   return [{
     ...$json,
   
     // Clean structured output
     output,
   
     // For Linear (number)
     priority: safePriority,
   
     // For UI / GitHub
     priorityLabel: priorityMap[safePriority],
     type: output.type,
     typeLabel: output.type,
   
     // Labels safe fallback
     labels: Array.isArray(output.labels) ? output.labels : []
   }];

3. Create task and post feedback

   Add a Linear Create an issue node: title uses the original issue title, teamId must be replaced with your actual Linear team ID, priorityId uses the formatted priority number, description uses the Merge node's description. Add a GitHub Create a comment on an issue node: operation set to createComment, issueNumber references the Github Trigger issue number, body uses a template to post classification results.

   Execution prompt · Source text。GitHub Create a comment on an issue node body field

   Copy prompt

   =Thanks for opening this issue!
   
   Our AI triage system classified this as:
   
   Type: {{ $('Code in JavaScript').item.json.typeLabel }}
   Priority: {{ $('Code in JavaScript').item.json.priorityLabel }}
   
   A Linear task has been created for the team.

You are setting up an n8n workflow to automatically process GitHub issues: trigger on new issue → AI classification → Linear task → GitHub comment. Please confirm you have configured credentials for GitHub, OpenAI, and Linear, and know the target repository and Linear teamId.