Streaming refactor

examples/basic_llm_call.py

@@ -21,7 +21,7 @@
 
 # imports
 from chuk_llm.llm.client import get_client
-from mcp_cli.llm.system_prompt_generator import SystemPromptGenerator
+from mcp_cli.chat.system_prompt import generate_system_prompt
 
 # load environment variables
 load_dotenv()
@@ -35,8 +35,8 @@ async def run_llm_diagnostic(provider: str, model: str, prompt: str) -> None:
     # get the client
     client = get_client(provider=provider, model=model)
 
-    # get the system prompt
-    system_prompt = SystemPromptGenerator().generate_prompt({})
+    # get the system prompt (tools are passed via API, not embedded in prompt)
+    system_prompt = generate_system_prompt()
     messages: List[Dict[str, Any]] = [
         {"role": "system", "content": system_prompt},
         {"role": "user", "content": prompt},

examples/mcp_round_trip.py

@@ -26,9 +26,9 @@
 from chuk_tool_processor.core.processor import ToolProcessor
 from chuk_tool_processor.models.tool_result import ToolResult
 
-# MCP CLI imports - only using llm_client and system_prompt_generator
+# MCP CLI imports
 from chuk_llm.llm.client import get_client
-from mcp_cli.llm.system_prompt_generator import SystemPromptGenerator
+from mcp_cli.chat.system_prompt import generate_system_prompt
 
 # Initialize colorama for colored output
 colorama_init(autoreset=True)
@@ -231,7 +231,8 @@ async def main() -> None:
 
         # 5) Send prompt to LLM
         client = get_client(provider=args.provider, model=args.model)
-        sys_prompt = SystemPromptGenerator().generate_prompt({"tools": openai_tools})
+        # Tools are passed via API, not embedded in system prompt
+        sys_prompt = generate_system_prompt(openai_tools)
         messages = [
             {"role": "system", "content": sys_prompt},
             {"role": "user", "content": args.prompt},

examples/mcp_round_trip_with_toolmanager.py

@@ -23,7 +23,7 @@
 # ── MCP & LLM helpers ───────────────────────────────────────────────────
 from chuk_llm.llm.client import get_client
 from mcp_cli.tools.manager import ToolManager
-from mcp_cli.llm.system_prompt_generator import SystemPromptGenerator
+from mcp_cli.chat.system_prompt import generate_system_prompt
 
 colorama_init(autoreset=True)
 
@@ -95,7 +95,8 @@ async def main() -> None:
 
         # 3️⃣  Initial LLM call (allow tool usage)
         client = get_client(provider=args.provider, model=args.model)
-        sys_prompt = SystemPromptGenerator().generate_prompt({"tools": llm_tools})
+        # Tools are passed via API, not embedded in system prompt
+        sys_prompt = generate_system_prompt(llm_tools)
         messages: List[Dict[str, str | None]] = [
             {"role": "system", "content": sys_prompt},
             {"role": "user", "content": args.prompt},

examples/ollama_llm_call.py

@@ -20,7 +20,7 @@
 
 # mcp cli imports
 from chuk_llm.llm.client import get_client
-from mcp_cli.llm.system_prompt_generator import SystemPromptGenerator
+from mcp_cli.chat.system_prompt import generate_system_prompt
 
 
 async def run_ollama_diagnostic(model: str, prompt: str) -> None:
@@ -30,7 +30,7 @@ async def run_ollama_diagnostic(model: str, prompt: str) -> None:
     except Exception as exc:
         sys.exit(f"[ERROR] Could not create Ollama client: {exc}")
 
-    system_prompt = SystemPromptGenerator().generate_prompt({})
+    system_prompt = generate_system_prompt()
     messages: List[Dict[str, Any]] = [
         {"role": "system", "content": system_prompt},
         {"role": "user", "content": prompt},