Self-improving AI Employees for everyone
Python ai-agentsai-employeeai-employeesassistantautonomous-agentsco-employeescompanygrok-botinstinctloop-agentsmuseown-your-datapersonalself-hosted
Fix pairing.restored event and reduce verbose logging
trohitgaafe9fc9f6a1
diff --git a/server/routers/whatsapp.py b/server/routers/whatsapp.pyindex d4b771ed..7ca2fdee 100644--- a/server/routers/whatsapp.py+++ b/server/routers/whatsapp.py@@ -107,7 +107,7 @@ class RPCClient:"""method = data.get("method", "")params = data.get("params", {})- logger.info(f"[WhatsApp RPC] Event: {method} params={params}")+ logger.debug(f"[WhatsApp RPC] Event: {method}")try:from services.status_broadcaster import get_status_broadcaster@@ -200,8 +200,7 @@ class RPCClient:await broadcaster.send_custom_event("whatsapp_message_sent", params)elif method == "event.message_received":- # Message received - broadcast as custom event- logger.info(f"[WhatsApp RPC] Broadcasting message_received event to trigger nodes")+ # Message received - broadcast as custom event for trigger nodesawait broadcaster.send_custom_event("whatsapp_message_received", params)# Forward to custom handler if setdiff --git a/server/services/ai.py b/server/services/ai.pyindex b62bf647..0f4e74a0 100644--- a/server/services/ai.py+++ b/server/services/ai.py@@ -255,41 +255,28 @@ def create_agent_node(chat_model):iteration = state.get("iteration", 0)max_iterations = state.get("max_iterations", 10)- logger.info(f"[LangGraph] Agent node invoked, iteration={iteration}, messages={len(messages)}")+ logger.debug(f"[LangGraph] Agent node invoked, iteration={iteration}, messages={len(messages)}")# Filter out messages with empty content using standardized utility# Prevents API errors/warnings from Gemini, Claude, and other providersfiltered_messages = filter_empty_messages(messages)if len(filtered_messages) != len(messages):- logger.info(f"[LangGraph] Filtered out {len(messages) - len(filtered_messages)} empty messages before LLM invoke")+ logger.debug(f"[LangGraph] Filtered out {len(messages) - len(filtered_messages)} empty messages")# Invoke the modelresponse = chat_model.invoke(filtered_messages)- logger.info(f"[LangGraph] LLM response type: {type(response)}, has tool_calls attr: {hasattr(response, 'tool_calls')}")-- # Debug log full response for troubleshooting empty responses (especially Gemini)- if hasattr(response, 'content'):- logger.info(f"[LangGraph] response.content type: {type(response.content)}, length: {len(str(response.content)) if response.content else 0}")+ logger.debug(f"[LangGraph] LLM response type: {type(response).__name__}")# Check for Gemini-specific response attributes (safety ratings, block reason)if hasattr(response, 'response_metadata'):meta = response.response_metadata- logger.info(f"[LangGraph] response_metadata: {meta}")- # Check for Gemini finish reason (could indicate blocked content)if meta.get('finish_reason') == 'SAFETY':logger.warning("[LangGraph] Gemini response blocked by safety filters")if meta.get('block_reason'):logger.warning(f"[LangGraph] Gemini block reason: {meta.get('block_reason')}")- # Log additional_kwargs which may contain Gemini-specific info- if hasattr(response, 'additional_kwargs') and response.additional_kwargs:- logger.info(f"[LangGraph] additional_kwargs: {response.additional_kwargs}")-- if hasattr(response, 'tool_calls'):- logger.info(f"[LangGraph] tool_calls value: {response.tool_calls}")-# Check for tool calls in the responsepending_tool_calls = []should_continue = False@@ -298,9 +285,7 @@ def create_agent_node(chat_model):# Model wants to use toolspending_tool_calls = response.tool_callsshould_continue = True- logger.info(f"[LangGraph] Agent requesting {len(pending_tool_calls)} tool call(s): {[tc.get('name', tc) for tc in pending_tool_calls]}")- else:- logger.info(f"[LangGraph] No tool calls in response, content preview: {str(response.content)[:100]}")+ logger.debug(f"[LangGraph] Agent requesting {len(pending_tool_calls)} tool call(s)")return {"messages": [response], # Will be appended via operator.add@@ -334,7 +319,7 @@ def create_tool_node(tool_executor: Callable):tool_args = tool_call.get("args", {})tool_id = tool_call.get("id", "")- logger.info(f"[LangGraph] Executing tool: {tool_name} with args: {tool_args}")+ logger.debug(f"[LangGraph] Executing tool: {tool_name}")try:# Directly await the async tool executor (proper async pattern)@@ -350,7 +335,7 @@ def create_tool_node(tool_executor: Callable):name=tool_name))- logger.info(f"[LangGraph] Tool {tool_name} completed with result: {result}")+ logger.debug(f"[LangGraph] Tool {tool_name} completed")return {"messages": tool_messages,@@ -395,7 +380,7 @@ def build_agent_graph(chat_model, tools: List = None, tool_executor: Callable =model_with_tools = chat_modelif tools:model_with_tools = chat_model.bind_tools(tools)- logger.info(f"[LangGraph] Bound {len(tools)} tools to model")+ logger.debug(f"[LangGraph] Bound {len(tools)} tools to model")# Add the agent nodeagent_fn = create_agent_node(model_with_tools)@@ -422,7 +407,7 @@ def build_agent_graph(chat_model, tools: List = None, tool_executor: Callable =# Tools always route back to agentgraph.add_edge("tools", "agent")- logger.info("[LangGraph] Built graph with tool execution loop")+ logger.debug("[LangGraph] Built graph with tool execution loop")else:# Simple graph without toolsgraph.add_conditional_edges(@@ -735,7 +720,7 @@ class AIService:temperature = float(flattened.get('temperature', 0.7))max_tokens = int(flattened.get('max_tokens') or flattened.get('maxTokens') or 1000)- logger.info(f"[LangGraph] AI Agent execution - Provider: {provider}, Model: {model}, Memory: {bool(memory_data)}, Tools: {len(tool_data) if tool_data else 0}")+ logger.info(f"[LangGraph] Agent: {provider}/{model}, tools={len(tool_data) if tool_data else 0}")# If no model specified or model doesn't match provider, use default from registryif not model or not is_model_valid_for_provider(model, provider):@@ -757,7 +742,7 @@ class AIService:})# Create LLM using the provider from node configuration- logger.info(f"[LangGraph] Creating {provider} model: {model}")+ logger.debug(f"[LangGraph] Creating {provider} model: {model}")chat_model = self.create_model(provider, api_key, model, temperature, max_tokens)# Build initial messages for state@@ -822,7 +807,7 @@ class AIService:tools.append(tool)tool_configs[tool.name] = config- logger.info(f"[LangGraph] Built {len(tools)} tools: {[t.name for t in tools]}")+ logger.debug(f"[LangGraph] Built {len(tools)} tools")# Create tool executor callbackasync def tool_executor(tool_name: str, tool_args: Dict) -> Any:@@ -832,7 +817,7 @@ class AIService:config = tool_configs.get(tool_name, {})tool_node_id = config.get('node_id')- logger.info(f"[LangGraph] tool_executor called for: {tool_name} (node_id: {tool_node_id})")+ logger.debug(f"[LangGraph] tool_executor: {tool_name}")# Broadcast executing status to the AI Agent nodeawait broadcast_status("executing_tool", {@@ -882,7 +867,7 @@ class AIService:})# Build and execute LangGraph agent- logger.info(f"[LangGraph] Building agent graph with {len(initial_messages)} initial messages and {len(tools)} tools")+ logger.debug(f"[LangGraph] Building agent graph with {len(initial_messages)} messages")agent_graph = build_agent_graph(chat_model,tools=tools if tools else None,@@ -925,7 +910,7 @@ class AIService:response_content = self._extract_text_content(raw_content, ai_response)iterations = final_state.get("iteration", 1)- logger.info(f"[LangGraph] Agent completed in {iterations} iteration(s), response length: {len(response_content)}")+ logger.info(f"[LangGraph] Agent completed in {iterations} iteration(s)")# Save to memory if connected (persist to database)# Only save non-empty messages using standardized validation@@ -1025,7 +1010,7 @@ class AIService:if db_schema:# Use database schema as source of truth- logger.info(f"[LangGraph] Using DB schema for tool node {node_id}")+ logger.debug(f"[LangGraph] Using DB schema for tool node {node_id}")tool_name = db_schema.get('tool_name', DEFAULT_TOOL_NAMES.get(node_type, f"tool_{node_label}"))tool_description = db_schema.get('tool_description', DEFAULT_TOOL_DESCRIPTIONS.get(node_type, f"Execute {node_label}"))# Use stored connected_services if available (for toolkit nodes)@@ -1082,7 +1067,7 @@ class AIService:'connected_services': connected_services # Pass through for execution}- logger.info(f"[LangGraph] Built tool '{tool_name}' from node type '{node_type}'")+ logger.debug(f"[LangGraph] Built tool '{tool_name}'")return tool, configexcept Exception as e:diff --git a/server/services/android/client.py b/server/services/android/client.pyindex dd7b2908..c4c2efd9 100644--- a/server/services/android/client.py+++ b/server/services/android/client.py@@ -314,6 +314,10 @@ class RelayWebSocketClient:if method == "pairing.connected":await self._handle_pairing_connected(params)+ elif method == "pairing.restored":+ # Handle auto-reconnect of previously paired device+ await self._handle_pairing_restored(params)+elif method == "pairing.disconnected":await self._handle_pairing_disconnected(params)@@ -345,6 +349,24 @@ class RelayWebSocketClient:if self.on_pairing_connected:await self.on_pairing_connected(params)+ async def _handle_pairing_restored(self, params: dict):+ """Handle pairing.restored event - auto-reconnect of previously paired device."""+ self.paired = True+ self.paired_device_id = params.get("device_id")+ self.paired_device_name = params.get("device_name")++ logger.info("[Relay] Android pairing restored (auto-reconnect)",+ device_id=self.paired_device_id,+ device_name=self.paired_device_name)++ await broadcast_connected(self.paired_device_id, self.paired_device_name)++ # Update saved session with latest info+ await self._save_pairing_session()++ if self.on_pairing_connected:+ await self.on_pairing_connected(params)+async def _save_pairing_session(self):"""Save pairing session to database for auto-reconnect."""try:diff --git a/server/services/deployment/manager.py b/server/services/deployment/manager.pyindex 4928f6c1..88a895d9 100644--- a/server/services/deployment/manager.py+++ b/server/services/deployment/manager.py@@ -495,7 +495,7 @@ class DeploymentManager:# Build filtered graphrun_filter = {trigger_node_id} | {n['id'] for n in downstream}- logger.info(f"[Run] run_filter has {len(run_filter)} nodes: {[n.get('type') for n in state.nodes if n['id'] in run_filter]}")+ logger.debug(f"[Run] run_filter has {len(run_filter)} nodes")filtered_nodes = []for node in state.nodes:@@ -511,7 +511,7 @@ class DeploymentManager:e for e in state.edgesif e.get('source') in run_filter and e.get('target') in run_filter]- logger.info(f"[Run] filtered_edges: {len(filtered_edges)} edges (from {len(state.edges)} total)")+ logger.debug(f"[Run] filtered_edges: {len(filtered_edges)} edges")# Execute filtered graph with deployment's workflow_id for scoped statusstatus_callback = self._status_callbacks.get(workflow_id)diff --git a/server/services/handlers/ai.py b/server/services/handlers/ai.pyindex d6f582b6..812293b7 100644--- a/server/services/handlers/ai.py+++ b/server/services/handlers/ai.py@@ -39,11 +39,7 @@ async def handle_ai_agent(memory_data = Nonetool_data: List[Dict[str, Any]] = [] # Collect connected tool nodes- # DEBUG: Log what we're working with- logger.info(f"[AI Agent] Processing node {node_id}, context has {len(edges) if edges else 0} edges, {len(nodes) if nodes else 0} nodes")- if edges:- toolkit_edges = [e for e in edges if 'androidTool' in str(e)]- logger.info(f"[AI Agent] Toolkit-related edges: {toolkit_edges}")+ logger.debug(f"[AI Agent] Processing node {node_id}, edges={len(edges) if edges else 0}, nodes={len(nodes) if nodes else 0}")if edges and nodes:for edge in edges:@@ -69,8 +65,7 @@ async def handle_ai_agent('memory_type': memory_type,'window_size': window_size}- logger.info("AI Agent connected memory node", node_id=node_id, execution_id=execution_id,- memory_session=memory_session_id, memory_type=memory_type)+ logger.debug("AI Agent connected memory node", memory_session=memory_session_id)# Tool detection (new) - any node connected to input-tools becomes a toolelif target_handle == 'input-tools':@@ -90,15 +85,6 @@ async def handle_ai_agent(if tool_type == 'androidTool':connected_services = []- # Debug: log all edges to understand the structure- logger.info(f"[Android Toolkit] Looking for edges targeting toolkit node: {source_node_id}")- logger.info(f"[Android Toolkit] Total edges in context: {len(edges)}")- for i, e in enumerate(edges):- logger.info(f"[Android Toolkit] Edge {i}: source={e.get('source')}, target={e.get('target')}, sourceHandle={e.get('sourceHandle')}, targetHandle={e.get('targetHandle')}")-- toolkit_edges = [e for e in edges if e.get('target') == source_node_id]- logger.info(f"[Android Toolkit] Edges targeting toolkit {source_node_id}: {toolkit_edges}")-# Scan edges for Android nodes connected to this toolkitfor service_edge in edges:# Skip if not targeting this androidTool node@@ -127,12 +113,10 @@ async def handle_ai_agent(logger.debug(f"Android toolkit connected service: {android_params.get('service_id')}")tool_entry['connected_services'] = connected_services- logger.info(f"Android toolkit has {len(connected_services)} connected services",- node_id=source_node_id, services=[s.get('service_id') for s in connected_services])+ logger.debug(f"Android toolkit has {len(connected_services)} connected services")tool_data.append(tool_entry)- logger.info("AI Agent connected tool node", node_id=node_id, execution_id=execution_id,- tool_type=tool_type, tool_node_id=source_node_id)+ logger.debug(f"AI Agent connected tool: {tool_type}")# Get broadcaster for real-time status updatesfrom services.status_broadcaster import get_status_broadcasterdiff --git a/server/services/parameter_resolver.py b/server/services/parameter_resolver.pyindex 6758a15c..2b3abd2a 100644--- a/server/services/parameter_resolver.py+++ b/server/services/parameter_resolver.py@@ -131,10 +131,11 @@ class ParameterResolver:# Case-insensitive lookupdata_lower = {k.lower(): v for k, v in connected_data.items()}- # Log template resolution context- template_params = {k: v for k, v in parameters.items() if isinstance(v, str) and '{{' in v}- if template_params:- logger.info(f"[ParameterResolver] Resolving templates: params_with_templates={list(template_params.keys())}, available_nodes={list(data_lower.keys())}")+ # Log template resolution context at debug level+ if logger.isEnabledFor(10): # DEBUG level+ template_params = {k: v for k, v in parameters.items() if isinstance(v, str) and '{{' in v}+ if template_params:+ logger.debug(f"[ParameterResolver] Resolving templates: {list(template_params.keys())}")def resolve(value: Any) -> Any:if isinstance(value, str) and '{{' in value: