search
User
account_tree AI Pipeline Dashboard

Multi-Agent AI System

Researcher → Designer → Validator → Debugger

กำลังดู: v2.0 Balanced QA
ACTIVE Creative Standard
settings ไปที่ Config
3
Pipeline Steps
3
Enabled Steps
69
Components
2
Design Rules
3
Forbidden
0
Req วันนี้
schema System Architecture Flow — ขั้นตอนการทำงาน end-to-end
person
User Builder / API
API
Node.js API Express.js Hub
settings_suggest
Config Engine Admin → API sync
play_circle
Pipeline Runner pipeline.js
output
Craft.js JSON Page Builder
Pipeline Agents
1 ON
auto_awesome
Claude Designer
Designer
CLAUDE
badge 🚦 AI Gate
🗂 design_patterns
📏 design_rules
💡 role_few_shot_examples
2 ON
rate_review
Strict QA
Reviewer
OPENAI
badge 🎬 AI Director
📏 design_rules
🚫 forbidden_patterns
💡 role_negative_examples
3 ON
verified
Validator
Validator
Local
🚫 forbidden_patterns
computerbigzwebcenter.live
→ Page Builder
POST /api/v2/ai/generate-page/stream hub api.bigzwebcenter.live (Node.js) SSE Stream
settings_suggest configEngine.js → fetch & cache pipeline config
GET admin.bigzwebcenter.live/api/v1/ai-engine/active-config
🔑 api_keys — gemini · openai · claude · serper · unsplash
🗂 pipeline_steps[] — step config array (type, model, provider, role_prompt…)
📏 design_rules — layout + typography constraints
🚫 forbidden_patterns — disallowed component patterns
🗃️ design_patterns[] — DesignPattern blueprints
play_circle pipeline.js → execute steps sequentially
1 auto_awesome Claude Designer ai_generate CLAUDE
Context APIs injected:
🗂 design_patterns 📏 design_rules 💡 role_few_shot_examples
GET /api/v1/ai-engine/active-config (Role: AI Gate)
AI call (primary):
POST api.anthropic.com/v1/messages
info ไม่มี fallback — ตั้งค่าใน Pipeline Builder
Output → next step: ✨ Craft.js Draft — JSON page nodes (ROOT + components)
↪ SSE emit: event: step   data: { step:"Claude Designer", status:"done", output_preview, duration_ms }
2 rate_review Strict QA ai_review OPENAI
Context APIs injected:
📏 design_rules 🚫 forbidden_patterns 💡 role_negative_examples
GET /api/v1/ai-engine/active-config (Role: AI Director)
AI call (primary):
POST api.openai.com/v1/chat/completions
info ไม่มี fallback — ตั้งค่าใน Pipeline Builder
Output → next step: 🔍 Review Report — feedback + must_fix[] + pass
↪ SSE emit: event: step   data: { step:"Strict QA", status:"done", output_preview, duration_ms }
3 verified Validator validator LOCAL
Context APIs injected:
🚫 forbidden_patterns
GET /api/v1/ai-engine/active-config
Output → next step: ✅ Validation — valid: true/false + errors[]
↪ SSE emit: event: step   data: { step:"Validator", status:"done", output_preview, duration_ms }
output Craft.js JSON → Page Builder / bigzwebcenter.live
SSE event: complete   data: { data: { ROOT: {…}, nodeId: {…} }, meta: { duration_ms } }
Legend:
Internal API call
External API call
AI provider call
SSE event emitted
Context injected
biotech
Pipeline Test Lab
ทดสอบ pipeline แบบ real-time — Timeline · Step Response · Timing Chart
POST /stream