Fixed ticket refining labels

.github/workflows/ticket-refinement.yml

@@ -2,7 +2,7 @@ name: Ticket refinement
 
 on:
     issues:
-        types: [opened, labeled]
+        types: [labeled]
 jobs:
     format:
         runs-on: ubuntu-latest

src/refine.mjs

@@ -52,7 +52,7 @@ what's happening now or the current state/gap
 
 What should happen instead
 
-## Steps to Reproduce
+## Steps to Reproduce / Desired Flow
 
 1. First step
 2. Second step
@@ -98,27 +98,36 @@ Implementation details, constraints, dependencies, design decisions
 			args: {type: {type: 'string', description: 'Ticket type', enum: ['Bug', 'DevOps', 'Document', 'Enhancement', 'Refactor', 'Security'], required: true}},
 			fn: (args) => type = args.type
 		}],
-		system: `You are a ticket formatter. Transform raw issue descriptions into structured tickets.
+		system: `Transform raw tickets into structured markdown following the template EXACTLY.
 
-**CRITICAL RULES:**
-1. Identify the ticket type (Bug, DevOps, Enhancement, Refactor, Security)
-2. Output MUST only contain the new ticket information in markdown, no extra fluff
-3. Follow the template structure EXACTLY:
-   - You must call the \`title\` tool EXACTLY ONCE with the title matching this format: [Module] - [Verb] [noun]
-     Example: Storage - Fix file uploads
-   - You must call the \`type\` tool EXACTLY ONCE with the identified ticket type
-   - Write a clear description
-   - For bugs: fill Steps to Reproduce with numbered list
-   - For enhancements/refactors: REMOVE the Steps to Reproduce section entirely
-   - For documentation: REMOVE the Current Behavior, Expected Behavior and Steps to Reproduce sections entirely
-   - Acceptance Criteria: convert requirements into checkboxes (- [ ])
-   - Weight scoring (0-5 each):
-     * Size: Number of modules, layers & files affected by change
-     * Complexity: Technical difficulty to implement
-     * Unknowns: Research/uncertainty in work estimation
-     * Calculate Total as sum of the three
-   - Remove sections that are not applicable based on ticket type
-   - Use proper markdown headers (##)
+**MANDATORY STEPS:**
+1. Identify ticket type: Bug, DevOps, Document, Enhancement, Refactor, or Security
+2. Call \`type\` tool EXACTLY ONCE with the type from step 1
+3. Call \`title\` tool EXACTLY ONCE in format: "[Module] - [Verb] [subject]"
+4. Output formatted markdown matching template structure below
+
+**TEMPLATE RULES:**
+- Use ## headers (match template exactly)
+- Description: Clear summary of the request
+- Current Behavior: What's happening now (remove for Document tickets)
+- Expected Behavior: What should happen (remove for Document tickets)
+- Steps to Reproduce: Numbered list for bugs, flow for enhancements, remove if not applicable
+- Additional Context: Logs, screenshots, links provided by user
+- Acceptance Criteria: Convert to checkboxes (- [ ] format)
+- Technical Notes: Implementation approach, constraints, dependencies
+- Weight table (use exact format below):
+
+| Effort / Weight | Score    |
+|-----------------|----------|
+| Size            | 0-5      |
+| Complexity      | 0-5      |
+| Unknowns        | 0-5      |
+| **Total**       | **0-15** |
+
+**SCORING:**
+- Size: # of modules/layers/files changed
+- Complexity: Technical difficulty
+- Unknowns: Research/uncertainty needed
 
 **README:**
 \`\`\`markdown
@@ -130,8 +139,7 @@ ${readme.trim() || 'No README available'}
 ${template.trim()}
 \`\`\`
 
-Output ONLY the formatted ticket, no explanation.`
-	});
+Output ONLY markdown. No explanations, labels, or extra formatting.`});
 
 	const messages = await ai.language.ask(`Title: ${issueData.title}\n\nDescription:\n${issueData.body || 'No description provided'}`).catch(() => []);
 	const body = messages?.pop()?.content;
@@ -160,5 +168,5 @@ Output ONLY the formatted ticket, no explanation.`
 		body: `["Kind/${type[0].toUpperCase() + type.slice(1).toLowerCase()}"]`
 	})
 
-	console.log(body);
+	console.log(`Title: ${title}\nType: ${type}\nBody:\n${body}`);
 })();