@su-record/vibe 2.6.21 → 2.6.23

package/commands/vibe.review.md

@@ -96,10 +96,17 @@ security-review:
 | Quick iteration | ❌ Standard review |
 | API cost concerns | ❌ Standard review |
 
-### Tool Invocation (Race Mode)
+### Tool Invocation (Race Mode - GPT + Gemini in parallel via Bash)
 
 ```bash
-node -e "import('@su-record/vibe/lib/ReviewRace').then(r => r.raceReview({reviewType: 'security', code: 'CODE_HERE'}).then(res => console.log(r.formatRaceResult(res))))"
+# Cross-platform path
+VIBE_SCRIPTS="$(node -p "process.env.APPDATA || require('os').homedir() + '/.config'")/vibe/hooks/scripts"
+
+# GPT review (Bash tool call 1)
+node "$VIBE_SCRIPTS/llm-orchestrate.js" gpt orchestrate-json "Review this code for [REVIEW_TYPE]. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. Code: [CODE_HERE]"
+
+# Gemini review (Bash tool call 2 - run in parallel)
+node "$VIBE_SCRIPTS/llm-orchestrate.js" gemini orchestrate-json "Review this code for [REVIEW_TYPE]. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. Code: [CODE_HERE]"
 ```
 
 ## Priority System

package/commands/vibe.run.md

@@ -905,11 +905,17 @@ After all scenarios are implemented, **GPT and Gemini review in parallel with cr
 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 ```
 
-**Race Review Invocation:**
+**Race Review Invocation (GPT + Gemini in parallel via Bash):**
 
 ```bash
-# Via vibe tools
-node -e "import('@su-record/vibe/tools').then(t => t.raceReview({reviewType: 'security', code: '[code]'}).then(r => console.log(t.formatRaceResult(r))))"
+# Cross-platform path
+VIBE_SCRIPTS="$(node -p "process.env.APPDATA || require('os').homedir() + '/.config'")/vibe/hooks/scripts"
+
+# GPT review (Bash tool call 1)
+node "$VIBE_SCRIPTS/llm-orchestrate.js" gpt orchestrate-json "Review this code for security, performance, and best practices. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. Code: [CODE]"
+
+# Gemini review (Bash tool call 2 - run in parallel)
+node "$VIBE_SCRIPTS/llm-orchestrate.js" gemini orchestrate-json "Review this code for security, performance, and best practices. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. Code: [CODE]"
 ```
 
 **Confidence-based Priority:**

package/commands/vibe.spec.review.md

@@ -165,48 +165,55 @@ Score: 96/100 ✅ PASSED
 
 ## Step 3: Race Review (GPT + Gemini Cross-Validation) - 3 Rounds (v2.6.9)
 
-**🚨 CRITICAL: Execute ALL 3 rounds with cross-validation. DO NOT skip.**
+**🚨🚨🚨 CRITICAL: YOU MUST EXECUTE ALL 3 ROUNDS. DO NOT SKIP THIS STEP. 🚨🚨🚨**
+
+**🚨 ABSOLUTE RULES FOR RACE REVIEW:**
+
+1. **YOU MUST** use the Bash tool to call `llm-orchestrate.js` directly
+2. **DO NOT** skip GPT/Gemini calls
+3. **DO NOT** simulate or fake review results
+4. **YOU MUST** run all 3 rounds sequentially (each round uses updated SPEC)
 
 > Race Mode reviews SPEC with both GPT and Gemini in parallel, then cross-validates findings for higher confidence.
 
-### 3.1 Race Review Invocation
+### 3.1 Review Loop (3 Rounds)
 
-```bash
-# Via vibe tools (recommended)
-node -e "import('@su-record/vibe/tools').then(t => t.raceReview({reviewType: 'general', code: '[SPEC content]', context: 'SPEC review round N/3'}).then(r => console.log(t.formatRaceResult(r))))"
-```
+**For EACH round (1, 2, 3), run GPT + Gemini in PARALLEL via Bash tool:**
 
-### 3.2 Review Loop (3 Rounds with Cross-Validation)
+**🚨 MANDATORY: Copy the EXACT path below. DO NOT modify or use alternative paths.**
 
-For each round (1 to 3):
+```bash
+# Cross-platform path (works on Windows/macOS/Linux)
+# ⚠️ COPY THIS EXACTLY - DO NOT USE ~/.claude/ or any other path!
+VIBE_SCRIPTS="$(node -p "process.env.APPDATA || require('os').homedir() + '/.config'")/vibe/hooks/scripts"
 
-**Run GPT and Gemini in PARALLEL with cross-validation:**
+# GPT review (Bash tool call 1)
+node "$VIBE_SCRIPTS/llm-orchestrate.js" gpt orchestrate-json "Review this SPEC for completeness, specificity, testability, security, and performance. Round [N]/3. Find issues and improvements. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. SPEC content: [SPEC_CONTENT]"
 
-```javascript
-// Race review - both models run simultaneously
-import('@su-record/vibe/tools').then(async t => {
-  const result = await t.raceReview({
-    reviewType: 'general',
-    code: specContent,
-    context: `SPEC review for ${featureName}. Stack: ${stack}. Round ${N}/3.`
-  });
-  console.log(t.formatRaceResult(result));
-});
+# Gemini review (Bash tool call 2 - run in parallel with GPT)
+node "$VIBE_SCRIPTS/llm-orchestrate.js" gemini orchestrate-json "Review this SPEC for completeness, specificity, testability, security, and performance. Round [N]/3. Find issues and improvements. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. SPEC content: [SPEC_CONTENT]"
 ```
 
-**Cross-validation rules:**
+**🚨 MANDATORY: Use the Bash tool to run BOTH commands above for EACH round.**
+**🚨 Run GPT and Gemini calls in PARALLEL (two separate Bash tool calls at once).**
+
+- Round 1: Run GPT + Gemini in parallel → Cross-validate → Apply fixes → Update SPEC file
+- Round 2: Run with updated SPEC → Cross-validate → Apply fixes → Update SPEC file
+- Round 3: Run with final SPEC → Cross-validate → Confirm no issues remain
+
+### 3.2 Cross-Validation Rules
 
 | Agreement | Priority | Action |
 |-----------|----------|--------|
-| Both agree (100%) | P1 | Auto-apply immediately |
-| One model (50%) | P2 | Auto-apply with note |
+| Both GPT + Gemini agree (100%) | P1 | Auto-apply immediately |
+| One model only (50%) | P2 | Auto-apply with note |
 
 **After each round:**
 
 1. Cross-validate findings (issues found by both → P1, single model → P2)
 2. Merge feedback with confidence scores
-3. Auto-apply P1/P2 improvements to SPEC and Feature files
-4. Continue to next round
+3. Auto-apply P1/P2 improvements to SPEC and Feature files (use Edit tool)
+4. Continue to next round with updated SPEC content
 
 **Output format:**
 ```

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@su-record/vibe",
-  "version": "2.6.21",
+  "version": "2.6.23",
   "description": "Vibe - Claude Code exclusive SPEC-driven AI coding framework with 35+ integrated tools",
   "type": "module",
   "main": "dist/cli/index.js",