fix(v9): Process ALL issues including EXISTING_REST with caching

Previously I incorrectly filtered out EXISTING_REST issues entirely.
The correct approach is to process ALL issues but use caching to
avoid duplicate AI calls:

1. Group ALL issues by rule:tool:language
2. Generate AI fix for first issue in each group
3. Apply cached fix to all issues in same group
4. EXISTING_REST issues with same rule as NEW issues share the fix

This ensures:
- All issues get fixes (not just actionable ones)
- AI calls are minimized through caching
- Re-runs use cached patterns from KB

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Author: alpsla

packages/agents/src/two-branch/analyzers/v9-report-formatter.ts

@@ -663,18 +663,12 @@ ${Object.entries(categoryCounts).map(([cat, count]) =>
   }
 
   private async formatIssueWithEducation(issue: Issue, severity: string): Promise<string> {
-    // SESSION 92: Use pre-generated fix if available, skip AI for EXISTING_REST
+    // SESSION 92 FIX: Use pre-generated fix if available from enrichWithAI cache
+    // All issues (including EXISTING_REST) should have fixes
     let fixSuggestion = (issue as any).fixSuggestion;
 
-    // Only generate new fix if:
-    // 1. No pre-existing fix, AND
-    // 2. Issue is actionable (NEW or EXISTING_MODIFIED based on category)
-    // Note: category is set during two-branch comparison and indicates actionability
-    const issueCategory = (issue as any).category as string | undefined;
-    const isActionable = issueCategory !== 'EXISTING_REST' && issueCategory !== 'RESOLVED' &&
-      (issueCategory === 'NEW' || issueCategory === 'EXISTING_MODIFIED' || !issueCategory);
-
-    if (!fixSuggestion && isActionable) {
+    // Generate fix only if not already cached from enrichWithAI
+    if (!fixSuggestion) {
       fixSuggestion = await this.generateDynamicFix(issue);
     }
 

packages/agents/src/two-branch/api/v9-analysis-service.ts

@@ -657,19 +657,18 @@ export class V9AnalysisService {
   ): Promise<EnrichedIssue[]> {
     console.log(`🤖 Step 4: AI Enrichment\n`);
 
-    // SESSION 92: Only enrich actionable issues (NEW + EXISTING_MODIFIED)
-    // Skip EXISTING_REST as they exist unchanged in both branches - not relevant for PR review
-    const actionableIssues = issues.filter(i =>
-      i.category === 'NEW' || i.category === 'EXISTING_MODIFIED'
-    );
-    const skippedCount = issues.length - actionableIssues.length;
-
-    if (skippedCount > 0) {
-      console.log(`   ℹ️  Skipping ${skippedCount} EXISTING_REST issues (not relevant for PR review)`);
-    }
+    // SESSION 92 FIX: Process ALL issues (including EXISTING_REST)
+    // Use caching to avoid duplicate AI calls - same rule gets same fix
+    const categoryCounts = {
+      NEW: issues.filter(i => i.category === 'NEW').length,
+      EXISTING_MODIFIED: issues.filter(i => i.category === 'EXISTING_MODIFIED').length,
+      EXISTING_REST: issues.filter(i => i.category === 'EXISTING_REST').length,
+      RESOLVED: issues.filter(i => i.category === 'RESOLVED').length
+    };
+    console.log(`   📊 Issues by category: NEW=${categoryCounts.NEW}, MODIFIED=${categoryCounts.EXISTING_MODIFIED}, EXISTING=${categoryCounts.EXISTING_REST}, RESOLVED=${categoryCounts.RESOLVED}`);
 
-    // Group only actionable issues to reduce AI calls
-    const groupingResult = groupIssues(actionableIssues);
+    // Group ALL issues to reduce AI calls - same rule:tool gets same fix
+    const groupingResult = groupIssues(issues);
     console.log(generateGroupingSummary(groupingResult));
 
     const { analyzed: priorityGroups } = prioritizeGroups(groupingResult.groups, maxAnalysis);