feat(agent): add comprehensive logging to LLM and tool execution nodes

- Add info and debug logging to llm_call node for tracking LLM invocations and configuration
- Add error handling and logging to tool_node for tool execution failures
- Add startup logging to stream_chat method with session details
- Improve observability of agent workflow and debugging capabilities

gns3server/agent/gns3_copilot/agent/gns3_copilot.py

@@ -128,6 +128,8 @@ class MessagesState(TypedDict):
 def llm_call(state: dict, config: RunnableConfig | None = None):
     """LLM decides whether to call a tool or not"""
 
+    logger.info("LLM call node invoked")
+
     # Get llm_config from request-scoped context variable
     from gns3server.agent.gns3_copilot.gns3_client import get_current_llm_config
     llm_config = get_current_llm_config()
@@ -140,9 +142,13 @@ def llm_call(state: dict, config: RunnableConfig | None = None):
             "topology_info": None,
         }
 
+    logger.debug("LLM config retrieved from context: provider=%s, model=%s",
+                 llm_config.get("provider"), llm_config.get("model"))
+
     # Defensive check: skip LLM call if no user messages
     messages = state.get("messages", [])
     if not messages or len(messages) == 0:
+        logger.warning("No messages in state, skipping LLM call")
         return {
             "messages": [],
             "llm_calls": state.get("llm_calls", 0),
@@ -216,13 +222,21 @@ def llm_call(state: dict, config: RunnableConfig | None = None):
     # print(full_messages)
 
     # Create fresh model with tools for each LLM call
+    logger.debug("Creating model with tools: provider=%s, model=%s",
+                 llm_config.get("provider"), llm_config.get("model"))
     model_with_tools = create_base_model_with_tools(
         tools,
         llm_config=llm_config
     )
 
+    logger.info("Invoking LLM with %d messages", len(full_messages))
+    response = model_with_tools.invoke(full_messages)
+
+    logger.info("LLM call completed: tool_calls=%d",
+                len(response.tool_calls) if hasattr(response, 'tool_calls') else 0)
+
     return {
-        "messages": [model_with_tools.invoke(full_messages)],
+        "messages": [response],
         "llm_calls": state.get("llm_calls", 0) + 1,
         "topology_info": topology_info,
     }
@@ -317,10 +331,21 @@ def generate_title(state: MessagesState, config: RunnableConfig | None = None) -
 def tool_node(state: dict, config: RunnableConfig | None = None):
     """Performs the tool call"""
 
+    tool_calls = state["messages"][-1].tool_calls
+    logger.info("Tool node invoked: tool_calls=%d", len(tool_calls))
+
     result = []
-    for tool_call in state["messages"][-1].tool_calls:
-        tool = tools_by_name[tool_call["name"]]
-        observation = tool.invoke(tool_call["args"])
+    for tool_call in tool_calls:
+        tool_name = tool_call["name"]
+        logger.debug("Executing tool: %s with args: %s", tool_name, tool_call["args"])
+        tool = tools_by_name[tool_name]
+        try:
+            observation = tool.invoke(tool_call["args"])
+            logger.debug("Tool %s completed: output_length=%d",
+                         tool_name, len(str(observation)) if observation else 0)
+        except Exception as e:
+            logger.error("Tool %s failed: %s", tool_name, e, exc_info=True)
+            observation = f"Error: {str(e)}"
         result.append(ToolMessage(
             content=observation,
             tool_call_id=tool_call["id"],

gns3server/agent/gns3_copilot/agent_service.py

@@ -123,13 +123,24 @@ class AgentService:
         Yields:
             Dict containing SSE-compatible response chunks
         """
+        log.info(
+            "Stream chat started: project_id=%s, user_id=%s, session_id=%s, mode=%s",
+            project_id,
+            user_id,
+            session_id,
+            mode,
+        )
+
         # Set request-scoped context variables (memory-only, not persisted)
         if jwt_token:
             from gns3server.agent.gns3_copilot.gns3_client import set_current_jwt_token
             set_current_jwt_token(jwt_token)
+            log.debug("JWT token set in context")
         if llm_config:
             from gns3server.agent.gns3_copilot.gns3_client import set_current_llm_config
             set_current_llm_config(llm_config)
+            log.debug("LLM config set in context: provider=%s, model=%s",
+                       llm_config.get("provider"), llm_config.get("model"))
 
         # Build config - only thread-safe identifiers
         config = {
@@ -152,12 +163,14 @@ class AgentService:
 
         # Get the compiled graph
         graph = await self._get_graph()
+        log.debug("LangGraph graph obtained, starting stream")
 
         # Stream events
         try:
             async for event in graph.astream_events(inputs, config=config, version="v2"):
                 chunk = self._convert_event_to_chunk(event, session_id)
                 if chunk:
+                    log.debug("Yielding chunk: type=%s", chunk.get("type"))
                     yield chunk
 
         except Exception as e:

gns3server/agent/gns3_copilot/gns3_client/connector_factory.py

@@ -30,6 +30,8 @@ import logging
 from typing import Optional
 from contextvars import ContextVar
 
+logger = logging.getLogger(__name__)
+
 # Context variables for request-scoped data
 # Automatically cleaned up when request context ends
 _jwt_token_context: ContextVar[Optional[str]] = ContextVar("_jwt_token_context", default=None)
@@ -39,27 +41,39 @@ _llm_config_context: ContextVar[Optional[dict]] = ContextVar("_llm_config_contex
 def set_current_jwt_token(token: str) -> None:
     """Set the JWT token for the current request context."""
     _jwt_token_context.set(token)
+    logger.debug("JWT token set in context")
 
 
 def get_current_jwt_token() -> Optional[str]:
     """Get the JWT token for the current request context."""
-    return _jwt_token_context.get()
+    token = _jwt_token_context.get()
+    if token:
+        logger.debug("JWT token retrieved from context")
+    else:
+        logger.warning("JWT token not found in context")
+    return token
 
 
 def set_current_llm_config(config: dict) -> None:
     """Set the LLM config for the current request context."""
     _llm_config_context.set(config)
+    logger.debug("LLM config set in context: provider=%s, model=%s",
+                config.get("provider"), config.get("model"))
 
 
 def get_current_llm_config() -> Optional[dict]:
     """Get the LLM config for the current request context."""
-    return _llm_config_context.get()
+    config = _llm_config_context.get()
+    if config:
+        logger.debug("LLM config retrieved from context: provider=%s, model=%s",
+                    config.get("provider"), config.get("model"))
+    else:
+        logger.warning("LLM config not found in context")
+    return config
 from uuid import UUID
 
 from gns3server.agent.gns3_copilot.gns3_client.custom_gns3fy import Gns3Connector
 
-logger = logging.getLogger(__name__)
-
 # Fallback default URL
 DEFAULT_GNS3_URL = "http://127.0.0.1:3080"
 
@@ -316,17 +330,22 @@ def _detect_url_for_api() -> Optional[str]:
     Returns:
         URL string, or None if detection failed
     """
+    logger.debug("Detecting GNS3 server URL for API calls")
+
     # Try Controller first
     url = _get_url_from_controller()
     if url:
+        logger.debug("Using URL from Controller for API call: %s", url)
         return url
 
     # Try Config
     url = _get_url_from_config()
     if url:
+        logger.debug("Using URL from Config for API call: %s", url)
         return url
 
     # Fallback
+    logger.debug("Using fallback URL for API call: %s", DEFAULT_GNS3_URL)
     return DEFAULT_GNS3_URL