Auto-Answer GitHub PR Tech Questions: GPT-4o + Notion + Slack n8n Workflow

GitHub PR comment trigger → detect question → GPT-4o generates answer → Notion archive → Slack notify

Who

Development teams and engineering leads using GitHub for code review

You get

When PR comments contain "how do I..." / "how to..." questions, automatically trigger AI to generate 2-3 line technical answers, store in Notion, and push to Slack

You need

n8n · GitHub · Notion · Google Sheets · Slack

Steps

1. Configure GitHub PR Comment Trigger

   Add a GitHub Trigger node in n8n, select event type pull_request_review_comment. Specify repository owner and name (template defaults to saurabhg97/weather-app, must be replaced). Configure OAuth2 authentication, ensuring the n8n instance can receive GitHub webhooks via a public URL. When triggered, captures comment body, file path, PR number, and author info.

2. Validate Webhook Data Integrity

   Add an IF node to check if incoming webhook data contains a valid comment URL. Condition: $json.body.comment.url is not empty. If validation passes, proceed to question detection branch; if failed, route error data to Google Sheets error log.

   Execution prompt · Source text。IF node conditions.leftValue: validate comment URL exists

   Copy prompt

   ={{ $json.body.comment.url }}

3. Detect Developer Question Pattern

   Add a second IF node, scanning comment body for "how do i" or "how to" keywords (case-insensitive). If matched, proceed to AI answer generation; if no match, workflow ends silently.

   Execution prompt · Source text。IF node conditions.leftValue: detect question keywords

   Copy prompt

   ={{ 
     (
       $json["body"]["comment"]["body"].toLowerCase().includes("how do i") ||
       $json["body"]["comment"]["body"].toLowerCase().includes("how to")
     ).toString()
   }}

4. Configure GPT-4o Model Connection

   Add an Azure OpenAI Chat Model node, select model gpt-4o. Configure Azure OpenAI API credentials (endpoint, API key, deployment name). This node serves as the language model for the subsequent AI Agent to call, and does not directly process input.

5. Generate AI Technical Answer

   Add an AI Agent node, connecting to the GPT-4o model from previous step. The node uses define prompt mode, injecting PR comment content and context into the prompt. System message sets answer style: short, clear, technically correct, maximum 2-3 lines.

   Execution prompt · Source text。AI Agent node parameters.text: dynamically inject comment content and PR context

   Copy prompt

   ={{ 
   `Question: ${$json["body"]["comment"]["body"]}
   
   Context: PR Link - ${$json["body"]["comment"]["html_url"]}
   Repository: ${$json["body"]["repository"]["full_name"]}
   ` 
   }}

   Execution prompt · Source text。AI Agent node parameters.options.systemMessage: system instruction controlling answer length and style

   Copy prompt

   =You are a helpful developer assistant. 
   Provide short, clear, and technically correct answers (2–3 lines max) to developer questions found in GitHub PR comments.

6. Extract Comment Metadata

   Add a Code node (JavaScript), extracting repository name, comment content, author username, file path, and PR number from the original webhook data, outputting as standard JSON format for the Notion node.

   Execution prompt · Source text。Code node parameters.jsCode: extract and structure GitHub comment metadata

   Copy prompt

   // Access the original webhook data
   const webhookData = $(' GitHub PR Comment Trigger').first().json;
   
   return [
     {
       json: {
         repository: webhookData.body.repository.full_name,
         comment: webhookData.body.comment.body,
         username: webhookData.body.comment.user.login,
         file_path: webhookData.body.comment.path,
         pr_number: webhookData.body.pull_request.number
       }
     }
   ];

7. Save to Notion Database

   Add a Notion node, select resource type databasePage. Specify database ID (template uses 'test db', must be replaced with actual database). Map fields from previous JSON step to Notion properties. **Note**: The template maps all fields to the same 'Name|title' key, which is likely a template error; actual use requires remapping to real database property names (e.g., Repository, Comment, Author, File Path, PR Number).

8. Push Answer to Slack

   Add a Slack node, select target user or channel (template hardcodes user U09HMPVD466, must be replaced). Message content combines AI-generated answer with original PR comment link for team members to quickly view context.

   Execution prompt · Source text。Slack node parameters.text: combine AI output with PR comment URL

   Copy prompt

   ={{ $('Generate AI Response for Developer Question').item.json.output }}{{ $('Detect Developer Question in PR Comment').item.json.body.comment.url }}

You are a setup assistant helping users replicate the "GitHub PR Auto-Answer" workflow in n8n. First confirm the user has ready: n8n instance (publicly accessible), GitHub OAuth credentials, Azure OpenAI GPT-4o access, Notion database, Slack workspace permissions, Google Sheets (for error logging). Then guide configuration step by step, with special attention to replacing hardcoded values in the template (repository name, Slack user, Notion database ID and property names).