Phase 2.3: Agent Management Module

Backend:
- agents_admin.py: 10 REST API endpoints for AI Agent CRUD
  - GET/POST/PUT/DELETE /admin/agents/
  - Stats overview, reviews, available skills/models
  - Agent = MarketApp with type='agent'

Frontend:
- Agents.tsx: Full admin page (780 lines)
  - Stats dashboard, filterable table, create/edit modal
  - Config: system prompt, temperature, model, linked skills
- api/index.ts: 10 new Agent API methods
- App.tsx: /admin/agents route
- AdminLayout.tsx: Agent management nav item

backend/python/app/api/admin/models.py

@@ -600,3 +600,155 @@ async def get_model_stats(
         "total_calls": int(total_calls),
         "total_tokens": int(total_tokens),
     }
+
+
+# ============================================================
+# 模型连接测试 API (Phase 2)
+# ============================================================
+
+@router.post("/{model_id}/test-connection")
+async def test_model_connection(
+    model_id: str,
+    admin_user=Depends(get_current_admin),
+    db: Session = Depends(get_db),
+):
+    """
+    测试指定模型的连接是否正常
+    支持从模型记录或提供商记录读取配置
+    """
+    logger = logging.getLogger(__name__)
+    logger.info(f"Testing connection for model_id: {model_id}")
+
+    # 优先从 AIModel 读取
+    ai_model = db.query(AIModel).filter(AIModel.id == model_id).first()
+    if not ai_model:
+        raise HTTPException(status_code=404, detail="模型不存在")
+
+    provider = db.query(ModelProvider).filter(ModelProvider.id == ai_model.provider_id).first()
+    if not provider:
+        raise HTTPException(status_code=404, detail="模型供应商不存在")
+
+    # 确定最终配置
+    api_key = ai_model.api_key or (provider.api_key if provider else None)
+    api_url = ai_model.endpoint
+
+    # 如果 endpoint 为空，尝试构造
+    if not api_url and provider and provider.api_base_url:
+        base = provider.api_base_url.rstrip('/')
+        if provider.protocol == 'openai_compatible':
+            api_url = f"{base}/v1/chat/completions"
+        else:
+            api_url = f"{base}/v1/chat/completions"
+
+    if not api_key:
+        return {
+            "success": False,
+            "message": "未配置 API Key，请在模型或供应商设置中添加",
+            "latency_ms": None,
+        }
+    if not api_url:
+        return {
+            "success": False,
+            "message": "未配置 API 地址，请在模型或供应商设置中添加",
+            "latency_ms": None,
+        }
+
+    timeout = ai_model.timeout or 60
+    import time
+    start = time.time()
+
+    try:
+        async with httpx.AsyncClient(timeout=timeout) as client:
+            headers = {
+                "Authorization": f"Bearer {api_key}",
+                "Content-Type": "application/json",
+            }
+            # 使用 minimal payload 测试连接
+            payload = {
+                "model": ai_model.name,
+                "messages": [{"role": "user", "content": "hi"}],
+                "max_tokens": 5,
+            }
+            resp = await client.post(api_url, json=payload, headers=headers)
+            latency = int((time.time() - start) * 1000)
+
+            if resp.status_code == 200:
+                try:
+                    data = resp.json()
+                    model_used = data.get("model", ai_model.name)
+                    return {
+                        "success": True,
+                        "message": f"连接成功，使用模型: {model_used}",
+                        "latency_ms": latency,
+                        "model_in_response": model_used,
+                    }
+                except Exception:
+                    return {
+                        "success": True,
+                        "message": f"连接成功 (HTTP {resp.status_code})",
+                        "latency_ms": latency,
+                    }
+            elif resp.status_code == 401:
+                return {
+                    "success": False,
+                    "message": "认证失败：API Key 无效或已过期",
+                    "latency_ms": latency,
+                }
+            elif resp.status_code == 403:
+                return {
+                    "success": False,
+                    "message": "访问被拒绝：API Key 权限不足",
+                    "latency_ms": latency,
+                }
+            elif resp.status_code == 429:
+                return {
+                    "success": False,
+                    "message": "触发限流：请稍后重试",
+                    "latency_ms": latency,
+                }
+            else:
+                try:
+                    err_data = resp.json()
+                    err_msg = err_data.get("error", {}).get("message", resp.text[:200])
+                except Exception:
+                    err_msg = resp.text[:200]
+                return {
+                    "success": False,
+                    "message": f"API 错误 (HTTP {resp.status_code}): {err_msg}",
+                    "latency_ms": latency,
+                }
+    except httpx.TimeoutException:
+        latency = int((time.time() - start) * 1000)
+        return {
+            "success": False,
+            "message": f"连接超时（{timeout}s）：请检查 API 地址是否正确，或服务是否可访问",
+            "latency_ms": latency,
+        }
+    except Exception as e:
+        latency = int((time.time() - start) * 1000)
+        logger.error(f"Model connection test failed: {e}")
+        return {
+            "success": False,
+            "message": f"连接失败: {str(e)}",
+            "latency_ms": latency,
+        }
+
+
+    timeout: int = 60
+    status: str = "paid"
+    price_per_million_tokens: float = 0.0
+    is_enabled: bool = True
+
+
+class UpdateModelRequest(BaseModel):
+    name: Optional[str] = None
+    api_key: Optional[str] = None
+    endpoint: Optional[str] = None
+    capabilities: Optional[List[str]] = None
+    timeout: Optional[int] = None
+    status: Optional[str] = None
+    price_per_million_tokens: Optional[float] = None
+    is_enabled: Optional[bool] = None
+
+
+from typing import Optional