💬 支持用户自定义分析视角注入（user_context 参数）

run_analysis.py

@@ -2,9 +2,10 @@
 TradingAgents 分析脚本
 用途：对持仓标的（NVDA / VOO）进行多 Agent 分析
 用法：
-  python run_analysis.py NVDA        # 分析 NVDA，使用今日日期
+  python run_analysis.py NVDA                          # 分析 NVDA，使用今日日期
-  python run_analysis.py VOO         # 分析 VOO
+  python run_analysis.py VOO                           # 分析 VOO
-  python run_analysis.py NVDA 2025-03-01  # 分析指定日期
+  python run_analysis.py NVDA 2026-03-20               # 指定日期
+  python run_analysis.py NVDA 2026-03-20 "中东地缘冲突..."  # 注入自定义分析视角
 """
 
 import sys
@@ -23,7 +24,7 @@ config = DEFAULT_CONFIG.copy()
 config["llm_provider"] = "google"
 config["deep_think_llm"] = "gemini-2.5-flash"
 config["quick_think_llm"] = "gemini-2.5-flash"
-config["max_debate_rounds"] = 1   # 减少辩论轮次，降低单次请求量
+config["max_debate_rounds"] = 1
 config["max_risk_discuss_rounds"] = 1
 config["data_vendors"] = {
     "core_stock_apis": "yfinance",
@@ -35,12 +36,21 @@ config["data_vendors"] = {
 # ── 入参解析 ─────────────────────────────────────────────────────────────────
 ticker = sys.argv[1].upper() if len(sys.argv) > 1 else "NVDA"
 analysis_date = sys.argv[2] if len(sys.argv) > 2 else str(date.today())
+user_context = sys.argv[3] if len(sys.argv) > 3 else ""
+
+# 如果没有传入上下文，交互式询问
+if not user_context and sys.stdin.isatty():
+    print("\n💬 是否有自定义分析视角？（直接回车跳过）")
+    print("   例：当前主导美股的核心因素是中东地缘冲突，请重点考虑此风险")
+    user_context = input("   > ").strip()
 
 print(f"\n{'='*60}")
 print(f"🤖 TradingAgents 多 Agent 分析")
 print(f"   标的：{ticker}")
 print(f"   日期：{analysis_date}")
 print(f"   模型：Gemini 2.5 Flash")
+if user_context:
+    print(f"   用户视角：{user_context[:60]}{'...' if len(user_context) > 60 else ''}")
 print(f"{'='*60}\n")
 
 # ── 执行分析（带 retry）────────────────────────────────────────────────────────
@@ -53,8 +63,8 @@ for attempt in range(1, MAX_RETRIES + 1):
             debug=False,
             config=config,
         )
-        final_state, decision = ta.propagate(ticker, analysis_date)
+        final_state, decision = ta.propagate(ticker, analysis_date, user_context)
-        break  # 成功则退出重试
+        break
     except Exception as e:
         print(f"⚠️  第 {attempt} 次失败: {type(e).__name__}: {str(e)[:120]}")
         if attempt < MAX_RETRIES:
@@ -76,8 +86,10 @@ os.makedirs(output_dir, exist_ok=True)
 output_file = os.path.join(output_dir, f"{ticker}_{analysis_date}.txt")
 
 with open(output_file, "w", encoding="utf-8") as f:
-    f.write(f"标的：{ticker}\n日期：{analysis_date}\n最终决策：{decision}\n\n")
+    f.write(f"标的：{ticker}\n日期：{analysis_date}\n最终决策：{decision}\n")
-    f.write("="*60 + "\n")
+    if user_context:
+        f.write(f"用户视角：{user_context}\n")
+    f.write("\n" + "="*60 + "\n")
 
     f.write("【交易员决策报告】\n")
     f.write(str(final_state.get("trader_investment_plan", "N/A")))

tradingagents/graph/propagation.py

@@ -16,11 +16,16 @@ class Propagator:
         self.max_recur_limit = max_recur_limit
 
     def create_initial_state(
-        self, company_name: str, trade_date: str
+        self, company_name: str, trade_date: str, user_context: str = ""
     ) -> Dict[str, Any]:
         """Create the initial state for the agent graph."""
+        # 支持注入用户自定义分析视角
+        if user_context:
+            human_msg = f"{company_name}\n\n[用户补充视角]\n{user_context}"
+        else:
+            human_msg = company_name
         return {
-            "messages": [("human", company_name)],
+            "messages": [("human", human_msg)],
             "company_of_interest": company_name,
             "trade_date": str(trade_date),
             "investment_debate_state": InvestDebateState(

tradingagents/graph/trading_graph.py

@@ -186,14 +186,20 @@ class TradingAgentsGraph:
             ),
         }
 
-    def propagate(self, company_name, trade_date):
+    def propagate(self, company_name, trade_date, user_context: str = ""):
-        """Run the trading agents graph for a company on a specific date."""
+        """Run the trading agents graph for a company on a specific date.
+        
+        Args:
+            company_name: Stock ticker symbol
+            trade_date: Analysis date (YYYY-MM-DD)
+            user_context: Optional user-provided analysis perspective or focus areas
+        """
 
         self.ticker = company_name
 
         # Initialize state
         init_agent_state = self.propagator.create_initial_state(
-            company_name, trade_date
+            company_name, trade_date, user_context
         )
         args = self.propagator.get_graph_args()