diff --git a/cli/main.py b/cli/main.py
index d41013eb..305f2643 100644
--- a/cli/main.py
+++ b/cli/main.py
@@ -32,7 +32,7 @@ lang = get_lang()
 
 app = typer.Typer(
     name="TradingAgents",
-    help=lang["welcome"] + ": " + lang["framework_subtitle"],
+    help=get_lang("welcome") + ": " + get_lang("framework_subtitle"),
     add_completion=True,  # Enable shell completion
 )
 
@@ -46,22 +46,22 @@ class MessageBuffer:
         self.final_report = None  # Store the complete final report
         self.agent_status = {
             # Analyst Team
-            lang["market_analyst"]: lang["pending"],
-            lang["social_analyst"]: lang["pending"],
-            lang["news_analyst"]: lang["pending"],
-            lang["fundamentals_analyst"]: lang["pending"],
+            get_lang("market_analyst"): get_lang("pending"),
+            get_lang("social_analyst"): get_lang("pending"),
+            get_lang("news_analyst"): get_lang("pending"),
+            get_lang("fundamentals_analyst"): get_lang("pending"),
             # Research Team
-            lang["bull_researcher"]: lang["pending"],
-            lang["bear_researcher"]: lang["pending"],
-            lang["research_manager"]: lang["pending"],
+            get_lang("bull_researcher"): get_lang("pending"),
+            get_lang("bear_researcher"): get_lang("pending"),
+            get_lang("research_manager"): get_lang("pending"),
             # Trading Team
-            lang["trader"]: lang["pending"],
+            get_lang("trader"): get_lang("pending"),
             # Risk Management Team
-            lang["risky_analyst"]: lang["pending"],
-            lang["neutral_analyst"]: lang["pending"],
-            lang["safe_analyst"]: lang["pending"],
+            get_lang("risky_analyst"): get_lang("pending"),
+            get_lang("neutral_analyst"): get_lang("pending"),
+            get_lang("safe_analyst"): get_lang("pending"),
             # Portfolio Management Team
-            lang["portfolio_manager"]: lang["pending"],
+            get_lang("portfolio_manager"): get_lang("pending"),
         }
         self.current_agent = None
         self.report_sections = {
@@ -106,13 +106,13 @@ class MessageBuffer:
         if latest_section and latest_content:
             # Format the current section for display
             section_titles = {
-                "market_report": lang["market_report"],
-                "sentiment_report": lang["sentiment_report"],
-                "news_report": lang["news_report"],
-                "fundamentals_report": lang["fundamentals_report"],
-                "investment_plan": lang["investment_plan"],
-                "trader_investment_plan": lang["trader_investment_plan"],
-                "final_trade_decision": lang["final_trade_decision"],
+                "market_report": get_lang("market_report"),
+                "sentiment_report": get_lang("sentiment_report"),
+                "news_report": get_lang("news_report"),
+                "fundamentals_report": get_lang("fundamentals_report"),
+                "investment_plan": get_lang("investment_plan"),
+                "trader_investment_plan": get_lang("trader_investment_plan"),
+                "final_trade_decision": get_lang("final_trade_decision"),
             }
             self.current_report = (
                 f"### {section_titles[latest_section]}\n{latest_content}"
@@ -134,37 +134,37 @@ class MessageBuffer:
                 "fundamentals_report",
             ]
         ):
-            report_parts.append(f"## {lang['analyst_team']}")
+            report_parts.append(f"## {get_lang("analyst_team")}")
             if self.report_sections["market_report"]:
                 report_parts.append(
-                    f"### {lang['market_report']}\n{self.report_sections['market_report']}"
+                    f"### {get_lang("market_report")}\n{self.report_sections['market_report']}"
                 )
             if self.report_sections["sentiment_report"]:
                 report_parts.append(
-                    f"### {lang['sentiment_report']}\n{self.report_sections['sentiment_report']}"
+                    f"### {get_lang("sentiment_report")}\n{self.report_sections['sentiment_report']}"
                 )
             if self.report_sections["news_report"]:
                 report_parts.append(
-                    f"### {lang['news_report']}\n{self.report_sections['news_report']}"
+                    f"### {get_lang("news_report")}\n{self.report_sections['news_report']}"
                 )
             if self.report_sections["fundamentals_report"]:
                 report_parts.append(
-                    f"### {lang['fundamentals_report']}\n{self.report_sections['fundamentals_report']}"
+                    f"### {get_lang("fundamentals_report")}\n{self.report_sections['fundamentals_report']}"
                 )
 
         # Research Team Reports
         if self.report_sections["investment_plan"]:
-            report_parts.append(f"## {lang['research_team']}")
+            report_parts.append(f"## {get_lang("research_team")}")
             report_parts.append(f"{self.report_sections['investment_plan']}")
 
         # Trading Team Reports
         if self.report_sections["trader_investment_plan"]:
-            report_parts.append(f"## {lang['trading_team']}")
+            report_parts.append(f"## {get_lang("trading_team")}")
             report_parts.append(f"{self.report_sections['trader_investment_plan']}")
 
         # Portfolio Management Decision
         if self.report_sections["final_trade_decision"]:
-            report_parts.append(f"## {lang['portfolio_management']}")
+            report_parts.append(f"## {get_lang("portfolio_management")}")
             report_parts.append(f"{self.report_sections['final_trade_decision']}")
 
         self.final_report = "\n\n".join(report_parts) if report_parts else None
@@ -193,9 +193,9 @@ def update_display(layout, spinner_text=None):
     # Header with welcome message
     layout["header"].update(
         Panel(
-            f"[bold green]{lang['welcome']}[/bold green]\n"
+            f"[bold green]{get_lang("welcome")}[/bold green]\n"
             "[dim]© [Tauric Research](https://github.com/TauricResearch)[/dim]",
-            title=lang["welcome"],
+            title=get_lang("welcome"),
             border_style="green",
             padding=(1, 2),
             expand=True,
@@ -212,22 +212,22 @@ def update_display(layout, spinner_text=None):
         padding=(0, 2),  # Add horizontal padding
         expand=True,  # Make table expand to fill available space
     )
-    progress_table.add_column(lang["team"], style="cyan", justify="center", width=20)
-    progress_table.add_column(lang["agent"], style="green", justify="center", width=20)
-    progress_table.add_column(lang["status"], style="yellow", justify="center", width=20)
+    progress_table.add_column(get_lang("team"), style="cyan", justify="center", width=20)
+    progress_table.add_column(get_lang("agent"), style="green", justify="center", width=20)
+    progress_table.add_column(get_lang("status"), style="yellow", justify="center", width=20)
 
     # Group agents by team
     teams = {
-        lang["analyst_team"]: [
-            lang["market_analyst"],
-            lang["social_analyst"],
-            lang["news_analyst"],
-            lang["fundamentals_analyst"],
+        get_lang("analyst_team"): [
+            get_lang("market_analyst"),
+            get_lang("social_analyst"),
+            get_lang("news_analyst"),
+            get_lang("fundamentals_analyst"),
         ],
-        lang["research_team"]: [lang["bull_researcher"], lang["bear_researcher"], lang["research_manager"]],
-        lang["trading_team"]: [lang["trader"]],
-        lang["risk_management"]: [lang["risky_analyst"], lang["neutral_analyst"], lang["safe_analyst"]],
-        lang["portfolio_management"]: [lang["portfolio_manager"]],
+        get_lang("research_team"): [get_lang("bull_researcher"), get_lang("bear_researcher"), get_lang("research_manager")],
+        get_lang("trading_team"): [get_lang("trader")],
+        get_lang("risk_management"): [get_lang("risky_analyst"), get_lang("neutral_analyst"), get_lang("safe_analyst")],
+        get_lang("portfolio_management"): [get_lang("portfolio_manager")],
     }
 
     for team, agents in teams.items():
@@ -236,14 +236,14 @@ def update_display(layout, spinner_text=None):
         status = message_buffer.agent_status[first_agent]
         if status == "in_progress":
             spinner = Spinner(
-                "dots", text=f"[blue]{lang['in_progress']}[/blue]", style="bold cyan"
+                "dots", text=f"[blue]{get_lang("in_progress")}[/blue]", style="bold cyan"
             )
             status_cell = spinner
         else:
             status_color = {
-                lang["pending"]: "yellow",
-                lang["completed"]: "green",
-                lang["error"]: "red",
+                get_lang("pending"): "yellow",
+                get_lang("completed"): "green",
+                get_lang("error"): "red",
             }.get(status, "white")
             status_cell = f"[{status_color}]{status}[/{status_color}]"
         progress_table.add_row(team, first_agent, status_cell)
@@ -253,14 +253,14 @@ def update_display(layout, spinner_text=None):
             status = message_buffer.agent_status[agent]
             if status == "in_progress":
                 spinner = Spinner(
-                    "dots", text=f"[blue]{lang['in_progress']}[/blue]", style="bold cyan"
+                    "dots", text=f"[blue]{get_lang("in_progress")}[/blue]", style="bold cyan"
                 )
                 status_cell = spinner
             else:
                 status_color = {
-                    lang["pending"]: "yellow",
-                    lang["completed"]: "green",
-                    lang["error"]: "red",
+                    get_lang("pending"): "yellow",
+                    get_lang("completed"): "green",
+                    get_lang("error"): "red",
                 }.get(status, "white")
                 status_cell = f"[{status_color}]{status}[/{status_color}]"
             progress_table.add_row("", agent, status_cell)
@@ -269,7 +269,7 @@ def update_display(layout, spinner_text=None):
         progress_table.add_row("─" * 20, "─" * 20, "─" * 20, style="dim")
 
     layout["progress"].update(
-        Panel(progress_table, title=lang["progress"], border_style="cyan", padding=(1, 2))
+        Panel(progress_table, title=get_lang("progress"), border_style="cyan", padding=(1, 2))
     )
 
     # Messages panel showing recent messages and tool calls
@@ -284,7 +284,7 @@ def update_display(layout, spinner_text=None):
     )
     messages_table.add_column("Time", style="cyan", width=8, justify="center")
     messages_table.add_column("Type", style="green", width=10, justify="center")
-    messages_table.add_column(lang["content"] if "content" in lang else "Content", style="white", no_wrap=False, ratio=1)
+    messages_table.add_column(get_lang("content") if "content" in lang else "Content", style="white", no_wrap=False, ratio=1)
 
     # Combine tool calls and messages
     all_messages = []
@@ -346,7 +346,7 @@ def update_display(layout, spinner_text=None):
         )
 
     layout["messages"].update(
-        Panel(messages_table, title=lang["messages_tools"], border_style="blue")
+        Panel(messages_table, title=get_lang("messages_tools"), border_style="blue")
     )
 
     # Analysis panel showing current report
@@ -354,7 +354,7 @@ def update_display(layout, spinner_text=None):
         layout["analysis"].update(
             Panel(
                 Markdown(message_buffer.current_report),
-                title=lang["current_report"],
+                title=get_lang("current_report"),
                 border_style="green",
                 padding=(1, 2),
             )
@@ -363,7 +363,7 @@ def update_display(layout, spinner_text=None):
         layout["analysis"].update(
             Panel(
                 "[italic]Waiting for analysis report...[/italic]",
-                title=lang["current_report"],
+                title=get_lang("current_report"),
                 border_style="green",
                 padding=(1, 2),
             )
@@ -395,9 +395,9 @@ def get_user_selections():
 
     # Create welcome box content
     welcome_content = f"{welcome_ascii}\n"
-    welcome_content += f"[bold green]TradingAgents: {lang['framework_subtitle']} - CLI[/bold green]\n\n"
-    welcome_content += f"[bold]{lang['workflow_steps_title']}[/bold]\n"
-    welcome_content += lang['workflow_steps']
+    welcome_content += f"[bold green]TradingAgents: {get_lang("framework_subtitle")} - CLI[/bold green]\n\n"
+    welcome_content += f"[bold]{get_lang("workflow_steps_title")}[/bold]\n"
+    welcome_content += get_lang("workflow_steps")
     welcome_content += (
         "[dim]Built by [Tauric Research](https://github.com/TauricResearch)[/dim]"
     )
@@ -407,8 +407,8 @@ def get_user_selections():
         welcome_content,
         border_style="green",
         padding=(1, 2),
-        title= lang["welcome"],
-        subtitle=lang["framework_subtitle"],
+        title= get_lang("welcome"),
+        subtitle=get_lang("framework_subtitle"),
     )
     console.print(Align.center(welcome_box))
     console.print()  # Add a blank line after the welcome box
@@ -424,7 +424,7 @@ def get_user_selections():
     # Step 1: Ticker symbol
     console.print(
         create_question_box(
-            lang["step1_title"], lang["step1_prompt"], lang["default_ticker"]
+            get_lang("step1_title"), get_lang("step1_prompt"), get_lang("default_ticker")
         )
     )
     selected_ticker = get_ticker()
@@ -433,8 +433,8 @@ def get_user_selections():
     default_date = datetime.datetime.now().strftime("%Y-%m-%d")
     console.print(
         create_question_box(
-            lang["step2_title"],
-            lang["step2_prompt"],
+            get_lang("step2_title"),
+            get_lang("step2_prompt"),
             default_date,
         )
     )
@@ -443,7 +443,7 @@ def get_user_selections():
     # Step 3: Select analysts
     console.print(
         create_question_box(
-            lang["step3_title"], lang["step3_prompt"]
+            get_lang("step3_title"), get_lang("step3_prompt")
         )
     )
     selected_analysts = select_analysts()
@@ -454,7 +454,7 @@ def get_user_selections():
     # Step 4: Research depth
     console.print(
         create_question_box(
-            lang["step4_title"], lang["step4_prompt"]
+            get_lang("step4_title"), get_lang("step4_prompt")
         )
     )
     selected_research_depth = select_research_depth()
@@ -462,7 +462,7 @@ def get_user_selections():
     # Step 5: OpenAI backend
     console.print(
         create_question_box(
-            lang["step5_title"], lang["step5_prompt"]
+            get_lang("step5_title"), get_lang("step5_prompt")
         )
     )
     selected_llm_provider, backend_url = select_llm_provider()
@@ -470,7 +470,7 @@ def get_user_selections():
     # Step 6: Thinking agents
     console.print(
         create_question_box(
-            lang["step6_title"], lang["step6_prompt"]
+            get_lang("step6_title"), get_lang("step6_prompt")
         )
     )
     selected_shallow_thinker = select_shallow_thinking_agent(selected_llm_provider)
diff --git a/cli/utils.py b/cli/utils.py
index 0e786ad4..ebc4143e 100644
--- a/cli/utils.py
+++ b/cli/utils.py
@@ -17,8 +17,8 @@ ANALYST_ORDER = [
 def get_ticker() -> str:
     """Prompt the user to enter a ticker symbol."""
     ticker = questionary.text(
-        lang["step1_prompt"],
-        validate=lambda x: len(x.strip()) > 0 or lang["ticker_validate"],
+        get_lang("step1_prompt"),
+        validate=lambda x: len(x.strip()) > 0 or get_lang("ticker_validate"),
         style=questionary.Style(
             [
                 ("text", "fg:green"),
@@ -49,8 +49,8 @@ def get_analysis_date() -> str:
             return False
 
     date = questionary.text(
-        lang["step2_prompt"],
-        validate=lambda x: validate_date(x.strip()) or lang["date_validate"],
+        get_lang("step2_prompt"),
+        validate=lambda x: validate_date(x.strip()) or get_lang("date_validate"),
         style=questionary.Style(
             [
                 ("text", "fg:green"),
@@ -69,12 +69,12 @@ def get_analysis_date() -> str:
 def select_analysts() -> List[AnalystType]:
     """Select analysts using an interactive checkbox."""
     choices = questionary.checkbox(
-        lang["step3_prompt"],
+        get_lang("step3_prompt"),
         choices=[
             questionary.Choice(lang.get(display.replace(" ", "_").lower(), display), value=value) for display, value in ANALYST_ORDER
         ],
-        instruction=lang["analyst_instruction"],
-        validate=lambda x: len(x) > 0 or lang["analyst_validate"],
+        instruction=get_lang("analyst_instruction"),
+        validate=lambda x: len(x) > 0 or get_lang("analyst_validate"),
         style=questionary.Style(
             [
                 ("checkbox-selected", "fg:green"),
@@ -97,17 +97,17 @@ def select_research_depth() -> int:
 
     # Define research depth options with their corresponding values
     DEPTH_OPTIONS = [
-        (lang["depth_shallow"], 1),
-        (lang["depth_medium"], 3),
-        (lang["depth_deep"], 5),
+        (get_lang("depth_shallow"), 1),
+        (get_lang("depth_medium"), 3),
+        (get_lang("depth_deep"), 5),
     ]
 
     choice = questionary.select(
-        lang["step4_prompt"],
+        get_lang("step4_prompt"),
         choices=[
             questionary.Choice(display, value=value) for display, value in DEPTH_OPTIONS
         ],
-        instruction=lang["depth_instruction"],
+        instruction=get_lang("depth_instruction"),
         style=questionary.Style(
             [
                 ("selected", "fg:yellow noinherit"),
diff --git a/tradingagents/agents/analysts/fundamentals_analyst.py b/tradingagents/agents/analysts/fundamentals_analyst.py
index dc81c304..e3ed51c2 100644
--- a/tradingagents/agents/analysts/fundamentals_analyst.py
+++ b/tradingagents/agents/analysts/fundamentals_analyst.py
@@ -3,8 +3,6 @@ import time
 import json
 from tradingagents.i18n import get_prompts
 
-prompts = get_prompts()
-
 def create_fundamentals_analyst(llm, toolkit):
     def fundamentals_analyst_node(state):
         current_date = state["trade_date"]
@@ -23,14 +21,14 @@ def create_fundamentals_analyst(llm, toolkit):
             ]
 
         system_message = (
-            prompts["analysts"]["fundamentals_analyst"]["system_message"]
+            get_prompts("analysts", "fundamentals_analyst", "system_message")
         )
 
         prompt = ChatPromptTemplate.from_messages(
             [
                 (
                     "system",
-                    prompts["analysts"]["template"]
+                    get_prompts("analysts", "template")
                 ),
                 MessagesPlaceholder(variable_name="messages"),
             ]
diff --git a/tradingagents/agents/analysts/market_analyst.py b/tradingagents/agents/analysts/market_analyst.py
index 73eb82cc..9c240ea4 100644
--- a/tradingagents/agents/analysts/market_analyst.py
+++ b/tradingagents/agents/analysts/market_analyst.py
@@ -3,8 +3,6 @@ import time
 import json
 from tradingagents.i18n import get_prompts
 
-prompts = get_prompts()
-
 def create_market_analyst(llm, toolkit):
 
     def market_analyst_node(state):
@@ -24,14 +22,14 @@ def create_market_analyst(llm, toolkit):
             ]
 
         system_message = (
-            prompts["analysts"]["market_analyst"]["system_message"]
+            get_prompts("analysts", "market_analyst", "system_message")
         )
 
         prompt = ChatPromptTemplate.from_messages(
             [
                 (
                     "system",
-                    prompts["analysts"]["template"]
+                    get_prompts("analysts", "template")
                 ),
                 MessagesPlaceholder(variable_name="messages"),
             ]
diff --git a/tradingagents/agents/analysts/news_analyst.py b/tradingagents/agents/analysts/news_analyst.py
index af3574be..94e71bba 100644
--- a/tradingagents/agents/analysts/news_analyst.py
+++ b/tradingagents/agents/analysts/news_analyst.py
@@ -3,8 +3,6 @@ import time
 import json
 from tradingagents.i18n import get_prompts
 
-prompts = get_prompts()
-
 def create_news_analyst(llm, toolkit):
     def news_analyst_node(state):
         current_date = state["trade_date"]
@@ -20,14 +18,14 @@ def create_news_analyst(llm, toolkit):
             ]
 
         system_message = (
-            prompts["analysts"]["news_analyst"]["system_message"]
+            get_prompts("analysts", "news_analyst", "system_message")
         )
 
         prompt = ChatPromptTemplate.from_messages(
             [
                 (
                     "system",
-                    prompts["analysts"]["template"]
+                    get_prompts("analysts", "template")
                 ),
                 MessagesPlaceholder(variable_name="messages"),
             ]
diff --git a/tradingagents/agents/analysts/social_media_analyst.py b/tradingagents/agents/analysts/social_media_analyst.py
index 0419082d..b0d72c7e 100644
--- a/tradingagents/agents/analysts/social_media_analyst.py
+++ b/tradingagents/agents/analysts/social_media_analyst.py
@@ -3,8 +3,6 @@ import time
 import json
 from tradingagents.i18n import get_prompts
 
-prompts = get_prompts()
-
 def create_social_media_analyst(llm, toolkit):
     def social_media_analyst_node(state):
         current_date = state["trade_date"]
@@ -19,14 +17,14 @@ def create_social_media_analyst(llm, toolkit):
             ]
 
         system_message = (
-            prompts["analysts"]["social_media_analyst"]["system_message"]
+            get_prompts("analysts", "social_media_analyst", "system_message")
         )
 
         prompt = ChatPromptTemplate.from_messages(
             [
                 (
                     "system",
-                    prompts["analysts"]["template"]
+                    get_prompts("analysts", "template")
                 ),
                 MessagesPlaceholder(variable_name="messages"),
             ]
diff --git a/tradingagents/agents/managers/research_manager.py b/tradingagents/agents/managers/research_manager.py
index 3ff27747..7f2bd409 100644
--- a/tradingagents/agents/managers/research_manager.py
+++ b/tradingagents/agents/managers/research_manager.py
@@ -2,8 +2,6 @@ import time
 import json
 from tradingagents.i18n import get_prompts
 
-prompts = get_prompts()
-
 def create_research_manager(llm, memory):
     def research_manager_node(state) -> dict:
         history = state["investment_debate_state"].get("history", "")
@@ -21,7 +19,7 @@ def create_research_manager(llm, memory):
         for i, rec in enumerate(past_memories, 1):
             past_memory_str += rec["recommendation"] + "\n\n"
 
-        prompt = prompts["managers"]["research_manager"] \
+        prompt = get_prompts("managers", "research_manager") \
             .replace("{past_memory_str}", past_memory_str) \
             .replace("{history}", history)
         response = llm.invoke(prompt)
diff --git a/tradingagents/agents/managers/risk_manager.py b/tradingagents/agents/managers/risk_manager.py
index a4a658e0..b3340eec 100644
--- a/tradingagents/agents/managers/risk_manager.py
+++ b/tradingagents/agents/managers/risk_manager.py
@@ -2,8 +2,6 @@ import time
 import json
 from tradingagents.i18n import get_prompts
 
-prompts = get_prompts()
-
 def create_risk_manager(llm, memory):
     def risk_manager_node(state) -> dict:
 
@@ -24,7 +22,7 @@ def create_risk_manager(llm, memory):
         for i, rec in enumerate(past_memories, 1):
             past_memory_str += rec["recommendation"] + "\n\n"
 
-        prompt = prompts["managers"]["risk_manager"] \
+        prompt = get_prompts("managers", "risk_manager") \
             .replace("{trader_plan}", trader_plan) \
             .replace("{past_memory_str}", past_memory_str) \
             .replace("{history}", history)
diff --git a/tradingagents/agents/researchers/bear_researcher.py b/tradingagents/agents/researchers/bear_researcher.py
index cc8b27db..2967a9b7 100644
--- a/tradingagents/agents/researchers/bear_researcher.py
+++ b/tradingagents/agents/researchers/bear_researcher.py
@@ -3,8 +3,6 @@ import time
 import json
 from tradingagents.i18n import get_prompts
 
-prompts = get_prompts()
-
 def create_bear_researcher(llm, memory):
     def bear_node(state) -> dict:
         investment_debate_state = state["investment_debate_state"]
@@ -24,7 +22,7 @@ def create_bear_researcher(llm, memory):
         for i, rec in enumerate(past_memories, 1):
             past_memory_str += rec["recommendation"] + "\n\n"
 
-        prompt = prompts["researchers"]["bear_researcher"] \
+        prompt = get_prompts("researchers", "bear_researcher") \
             .replace("{market_research_report}", market_research_report) \
             .replace("{sentiment_report}", sentiment_report) \
             .replace("{news_report}", news_report) \
diff --git a/tradingagents/agents/researchers/bull_researcher.py b/tradingagents/agents/researchers/bull_researcher.py
index 37cbc7b4..e05b09b1 100644
--- a/tradingagents/agents/researchers/bull_researcher.py
+++ b/tradingagents/agents/researchers/bull_researcher.py
@@ -3,8 +3,6 @@ import time
 import json
 from tradingagents.i18n import get_prompts
 
-prompts = get_prompts()
-
 def create_bull_researcher(llm, memory):
     def bull_node(state) -> dict:
         investment_debate_state = state["investment_debate_state"]
@@ -24,7 +22,7 @@ def create_bull_researcher(llm, memory):
         for i, rec in enumerate(past_memories, 1):
             past_memory_str += rec["recommendation"] + "\n\n"
 
-        prompt = prompts["researchers"]["bull_researcher"] \
+        prompt = get_prompts("researchers", "bull_researcher") \
             .replace("{market_research_report}", market_research_report) \
             .replace("{sentiment_report}", sentiment_report) \
             .replace("{news_report}", news_report) \
diff --git a/tradingagents/agents/risk_mgmt/aggresive_debator.py b/tradingagents/agents/risk_mgmt/aggresive_debator.py
index 7a9ea874..9457fb0d 100644
--- a/tradingagents/agents/risk_mgmt/aggresive_debator.py
+++ b/tradingagents/agents/risk_mgmt/aggresive_debator.py
@@ -2,8 +2,6 @@ import time
 import json
 from tradingagents.i18n import get_prompts
 
-prompts = get_prompts()
-
 def create_risky_debator(llm):
     def risky_node(state) -> dict:
         risk_debate_state = state["risk_debate_state"]
@@ -20,7 +18,7 @@ def create_risky_debator(llm):
 
         trader_decision = state["trader_investment_plan"]
 
-        prompt = prompts["risk_mgmt"]["aggressive_debator"] \
+        prompt = get_prompts("risk_mgmt", "aggressive_debator") \
             .replace("{trader_decision}", trader_decision) \
             .replace("{market_research_report}", market_research_report) \
             .replace("{sentiment_report}", sentiment_report) \
diff --git a/tradingagents/agents/risk_mgmt/conservative_debator.py b/tradingagents/agents/risk_mgmt/conservative_debator.py
index e16b434e..e3bfd7ce 100644
--- a/tradingagents/agents/risk_mgmt/conservative_debator.py
+++ b/tradingagents/agents/risk_mgmt/conservative_debator.py
@@ -3,8 +3,6 @@ import time
 import json
 from tradingagents.i18n import get_prompts
 
-prompts = get_prompts()
-
 def create_safe_debator(llm):
     def safe_node(state) -> dict:
         risk_debate_state = state["risk_debate_state"]
@@ -21,7 +19,7 @@ def create_safe_debator(llm):
 
         trader_decision = state["trader_investment_plan"]
 
-        prompt = prompts["risk_mgmt"]["conservative_debator"] \
+        prompt = get_prompts("risk_mgmt", "conservative_debator") \
             .replace("{trader_decision}", trader_decision) \
             .replace("{market_research_report}", market_research_report) \
             .replace("{sentiment_report}", sentiment_report) \
diff --git a/tradingagents/agents/risk_mgmt/neutral_debator.py b/tradingagents/agents/risk_mgmt/neutral_debator.py
index 2cea3175..96dc188d 100644
--- a/tradingagents/agents/risk_mgmt/neutral_debator.py
+++ b/tradingagents/agents/risk_mgmt/neutral_debator.py
@@ -2,8 +2,6 @@ import time
 import json
 from tradingagents.i18n import get_prompts
 
-prompts = get_prompts()
-
 def create_neutral_debator(llm):
     def neutral_node(state) -> dict:
         risk_debate_state = state["risk_debate_state"]
@@ -20,7 +18,7 @@ def create_neutral_debator(llm):
 
         trader_decision = state["trader_investment_plan"]
 
-        prompt = prompts["risk_mgmt"]["neutral_debator"] \
+        prompt = get_prompts("risk_mgmt", "neutral_debator") \
             .replace("{trader_decision}", trader_decision) \
             .replace("{market_research_report}", market_research_report) \
             .replace("{sentiment_report}", sentiment_report) \
diff --git a/tradingagents/agents/trader/trader.py b/tradingagents/agents/trader/trader.py
index f24dd26f..3e95054a 100644
--- a/tradingagents/agents/trader/trader.py
+++ b/tradingagents/agents/trader/trader.py
@@ -3,8 +3,6 @@ import time
 import json
 from tradingagents.i18n import get_prompts
 
-prompts = get_prompts()
-
 def create_trader(llm, memory):
     def trader_node(state, name):
         company_name = state["company_of_interest"]
@@ -23,7 +21,7 @@ def create_trader(llm, memory):
 
         context = {
             "role": "user",
-            "content": prompts["trader"]["user_message"] \
+            "content": get_prompts("trader", "user_message") \
                 .replace("{company_name}", company_name) \
                 .replace("{investment_plan}", investment_plan)
         }
@@ -31,7 +29,7 @@ def create_trader(llm, memory):
         messages = [
             {
                 "role": "system",
-                "content": prompts["trader"]["system_message"] \
+                "content": get_prompts("trader", "system_message") \
                     .replace("{past_memory_str}", past_memory_str),
             },
             context,
diff --git a/tradingagents/graph/reflection.py b/tradingagents/graph/reflection.py
index 456157c4..59a06f25 100644
--- a/tradingagents/graph/reflection.py
+++ b/tradingagents/graph/reflection.py
@@ -4,8 +4,6 @@ from typing import Dict, Any
 from langchain_openai import ChatOpenAI
 from tradingagents.i18n import get_prompts
 
-prompts = get_prompts()
-
 class Reflector:
     """Handles reflection on decisions and updating memory."""
 
@@ -16,7 +14,7 @@ class Reflector:
 
     def _get_reflection_prompt(self) -> str:
         """Get the system prompt for reflection."""
-        return prompts["reflection"]["system_message"]
+        return get_prompts("reflection", "system_message")
 
     def _extract_current_situation(self, current_state: Dict[str, Any]) -> str:
         """Extract the current market situation from the state."""
@@ -35,7 +33,7 @@ class Reflector:
             ("system", self.reflection_system_prompt),
             (
                 "human",
-                prompts["reflection"]["user_message"] \
+                get_prompts("reflection", "user_message") \
                     .replace("{returns_losses}", returns_losses) \
                     .replace("{report}", report) \
                     .replace("{situation}", situation)
diff --git a/tradingagents/graph/signal_processing.py b/tradingagents/graph/signal_processing.py
index b87cd630..0f2f23e5 100644
--- a/tradingagents/graph/signal_processing.py
+++ b/tradingagents/graph/signal_processing.py
@@ -3,8 +3,6 @@
 from langchain_openai import ChatOpenAI
 from tradingagents.i18n import get_prompts
 
-prompts = get_prompts()
-
 class SignalProcessor:
     """Processes trading signals to extract actionable decisions."""
 
@@ -25,7 +23,7 @@ class SignalProcessor:
         messages = [
             (
                 "system",
-                prompts["signal_processor"]["system_message"],
+                get_prompts("signal_processor", "system_message"),
             ),
             ("human", full_signal),
         ]
diff --git a/tradingagents/i18n/__init__.py b/tradingagents/i18n/__init__.py
index d00ff886..b6f8aa6d 100644
--- a/tradingagents/i18n/__init__.py
+++ b/tradingagents/i18n/__init__.py
@@ -1,22 +1,37 @@
+from functools import reduce
 import importlib
 from tradingagents.default_config import DEFAULT_CONFIG
 
-def get_lang():
+def get_value(dictionary: dict, *keys, default=None):
+    """
+    Get values from a dictionary using a list of keys.
+    If a key is not found, it returns the default value.
+    """
+    try:
+        return reduce((lambda d, key: d[key]), keys, dictionary)
+    except (KeyError, TypeError):
+        return default
+
+def get_lang(*keys, default="") -> str | dict:
     lang_code = DEFAULT_CONFIG.get("language", "zh")
     try:
         lang_module = importlib.import_module(f"tradingagents.i18n.{lang_code}")
-        return lang_module.LANG
+        if not keys:
+            return lang_module.LANG
+        return get_value(lang_module.LANG, *keys, default=default)
     except Exception:
         # fallback to zh
         from .interface.zh import LANG
-        return LANG 
+        if not keys:
+            return LANG
+        return get_value(LANG, *keys, default=default) 
     
-def get_prompts():
+def get_prompts(*keys, default="") -> str:
     lang_code = DEFAULT_CONFIG.get("language", "zh")
     try:
         lang_module = importlib.import_module(f"tradingagents.i18n.{lang_code}")
-        return lang_module.PROMPTS
+        return get_value(lang_module.PROMPTS, *keys, default=default)
     except Exception:
         # fallback to zh
         from .prompts.zh import PROMPTS
-        return PROMPTS
\ No newline at end of file
+        return get_value(PROMPTS, *keys, default=default)
\ No newline at end of file