fix(agent): show agent run steps, fixes #21718

api/core/workflow/nodes/agent/agent_node.py

@@ -1,4 +1,5 @@
 import json
+import uuid
 from collections.abc import Generator, Mapping, Sequence
 from typing import Any, Optional, cast
 
@@ -102,14 +103,46 @@ class AgentNode(ToolNode):
 
         try:
             # convert tool messages
+            agent_thoughts = []
+
+            from core.tools.entities.tool_entities import ToolInvokeMessage
+            
+            thought_log_message = ToolInvokeMessage(
+                type=ToolInvokeMessage.MessageType.LOG,
+                message=ToolInvokeMessage.LogMessage(
+                    id=str(uuid.uuid4()),
+                    label=f"Agent Strategy: {cast(AgentNodeData, self.node_data).agent_strategy_name}",
+                    parent_id=None,
+                    error=None,
+                    status=ToolInvokeMessage.LogMessage.LogStatus.START,
+                    data={
+                        "strategy": cast(AgentNodeData, self.node_data).agent_strategy_name,
+                        "parameters": parameters_for_log,
+                        "thought_process": "Agent strategy execution started",
+                    },
+                    metadata={
+                        "icon": self.agent_strategy_icon,
+                        "agent_strategy": cast(AgentNodeData, self.node_data).agent_strategy_name,
+                    },
+                ),
+            )
+
+            from core.tools.entities.tool_entities import ToolInvokeMessage
+            
+            def enhanced_message_stream():
+                
+                yield thought_log_message
+                
+                yield from message_stream
 
             yield from self._transform_message(
-                message_stream,
+                enhanced_message_stream(),
                 {
                     "icon": self.agent_strategy_icon,
                     "agent_strategy": cast(AgentNodeData, self.node_data).agent_strategy_name,
                 },
                 parameters_for_log,
+                agent_thoughts,
             )
         except PluginDaemonClientSideError as e:
             yield RunCompletedEvent(

api/core/workflow/nodes/tool/tool_node.py

@@ -1,5 +1,5 @@
 from collections.abc import Generator, Mapping, Sequence
-from typing import Any, cast
+from typing import Any, Optional, cast
 
 from sqlalchemy import select
 from sqlalchemy.orm import Session
@@ -188,6 +188,7 @@ class ToolNode(BaseNode[ToolNodeData]):
         messages: Generator[ToolInvokeMessage, None, None],
         tool_info: Mapping[str, Any],
         parameters_for_log: dict[str, Any],
+        agent_thoughts: Optional[list] = None,
     ) -> Generator:
         """
         Convert ToolInvokeMessages into tuple[plain_text, files]
@@ -365,10 +366,41 @@ class ToolNode(BaseNode[ToolNodeData]):
 
                 yield agent_log
 
+        # Add agent_logs to outputs['json'] to ensure frontend can access thinking process
+        json_output = json.copy()
+        if agent_logs:
+            if not json_output:
+                json_output = {}
+            elif isinstance(json_output, list) and len(json_output) == 1:
+                # If json is a list with only one element, convert it to a dictionary
+                json_output = json_output[0] if isinstance(json_output[0], dict) else {"data": json_output[0]}
+            elif isinstance(json_output, list):
+                # If json is a list with multiple elements, create a dictionary containing all data
+                json_output = {"data": json_output}
+            
+            # Ensure json_output is a dictionary type
+            if not isinstance(json_output, dict):
+                json_output = {"data": json_output}
+            
+            # Add agent_logs to json output
+            json_output["agent_logs"] = [
+                {
+                    "id": log.id,
+                    "parent_id": log.parent_id,
+                    "error": log.error,
+                    "status": log.status,
+                    "data": log.data,
+                    "label": log.label,
+                    "metadata": log.metadata,
+                    "node_id": log.node_id,
+                }
+                for log in agent_logs
+            ]
+
         yield RunCompletedEvent(
             run_result=NodeRunResult(
                 status=WorkflowNodeExecutionStatus.SUCCEEDED,
-                outputs={"text": text, "files": ArrayFileSegment(value=files), "json": json, **variables},
+                outputs={"text": text, "files": ArrayFileSegment(value=files), "json": json_output, **variables},
                 metadata={
                     **agent_execution_metadata,
                     WorkflowNodeExecutionMetadataKey.TOOL_INFO: tool_info,

api/services/enterprise/enterprise_service.py

@@ -29,7 +29,7 @@ class EnterpriseService:
             raise ValueError("No data found.")
         try:
             # parse the UTC timestamp from the response
-            return datetime.fromisoformat(data.replace("Z", "+00:00"))
+            return datetime.fromisoformat(data)
         except ValueError as e:
             raise ValueError(f"Invalid date format: {data}") from e
 
@@ -40,7 +40,7 @@ class EnterpriseService:
             raise ValueError("No data found.")
         try:
             # parse the UTC timestamp from the response
-            return datetime.fromisoformat(data.replace("Z", "+00:00"))
+            return datetime.fromisoformat(data)
         except ValueError as e:
             raise ValueError(f"Invalid date format: {data}") from e
 

api/tests/integration_tests/workflow/nodes/test_llm.py

@@ -119,11 +119,11 @@ def test_execute_llm(flask_req_ctx):
     mock_usage = LLMUsage(
         prompt_tokens=30,
         prompt_unit_price=Decimal("0.001"),
-        prompt_price_unit=Decimal("1000"),
+        prompt_price_unit=Decimal(1000),
         prompt_price=Decimal("0.00003"),
         completion_tokens=20,
         completion_unit_price=Decimal("0.002"),
-        completion_price_unit=Decimal("1000"),
+        completion_price_unit=Decimal(1000),
         completion_price=Decimal("0.00004"),
         total_tokens=50,
         total_price=Decimal("0.00007"),
@@ -222,11 +222,11 @@ def test_execute_llm_with_jinja2(flask_req_ctx, setup_code_executor_mock):
     mock_usage = LLMUsage(
         prompt_tokens=30,
         prompt_unit_price=Decimal("0.001"),
-        prompt_price_unit=Decimal("1000"),
+        prompt_price_unit=Decimal(1000),
         prompt_price=Decimal("0.00003"),
         completion_tokens=20,
         completion_unit_price=Decimal("0.002"),
-        completion_price_unit=Decimal("1000"),
+        completion_price_unit=Decimal(1000),
         completion_price=Decimal("0.00004"),
         total_tokens=50,
         total_price=Decimal("0.00007"),

core/prompt/entities/advanced_prompt_entities.py

@@ -0,0 +1,19 @@
+from typing import Optional, Any
+from pydantic import BaseModel, Field, model_validator
+from core.prompt.entities.role_prefix import RolePrefix
+from core.prompt.entities.window import Window
+
+class MemoryConfig(BaseModel):
+    role_prefix: RolePrefix = Field(default_factory=RolePrefix)
+    window: Window = Field(default_factory=Window)
+    memory_key: Optional[str] = Field(None)
+
+    # The `model_validate` method is used to create a `MemoryConfig` object from a dictionary.
+    @model_validator(mode="before")
+    @classmethod
+    def pre_validate(cls, values: Any) -> Any:
+        if "role_prefix" not in values:
+            values["role_prefix"] = {}
+        if "window" not in values:
+            values["window"] = {}
+        return values

web/.env.local.save

@@ -0,0 +1,60 @@
+\# For production release, change this to PRODUCTION
+NEXT_PUBLIC_DEPLOY_ENV=DEVELOPMENT
+# The deployment edition, SELF_HOSTED
+NEXT_PUBLIC_EDITION=SELF_HOSTED
+# The base URL of console application, refers to the Console base URL of WEB service if console domain is
+# different from api or web app domain.
+# example: http://cloud.dify.ai/console/api
+NEXT_PUBLIC_API_PREFIX=http://localhost:5001/console/api
+# The URL for Web APP, refers to the Web App base URL of WEB service if web app domain is different from
+# console or api domain.
+# example: http://udify.app/api
+NEXT_PUBLIC_PUBLIC_API_PREFIX=http://localhost:5001/api
+# The API PREFIX for MARKETPLACE
+NEXT_PUBLIC_MARKETPLACE_API_PREFIX=https://marketplace.dify.ai/api/v1
+# The URL for MARKETPLACE
+NEXT_PUBLIC_MARKETPLACE_URL_PREFIX=https://marketplace.dify.ai
+
+# SENTRY
+NEXT_PUBLIC_SENTRY_DSN=
+
+# Disable Next.js Telemetry (https://nextjs.org/telemetry)
+NEXT_TELEMETRY_DISABLED=1
+
+# Disable Upload Image as WebApp icon default is false
+NEXT_PUBLIC_UPLOAD_IMAGE_AS_ICON=false
+
+# The timeout for the text generation in millisecond
+NEXT_PUBLIC_TEXT_GENERATION_TIMEOUT_MS=60000
+
+# CSP https://developer.mozilla.org/en-US/docs/Web/HTTP/CSP
+NEXT_PUBLIC_CSP_WHITELIST=
+# Default is not allow to embed into iframe to prevent Clickjacking: https://owasp.org/www-community/attacks/Clickjacking
+NEXT_PUBLIC_ALLOW_EMBED=
+
+# Github Access Token, used for invoking Github API
+NEXT_PUBLIC_GITHUB_ACCESS_TOKEN=
+# The maximum number of top-k value for RAG.
+NEXT_PUBLIC_TOP_K_MAX_VALUE=10
+
+# The maximum number of tokens for segmentation
+NEXT_PUBLIC_INDEXING_MAX_SEGMENTATION_TOKENS_LENGTH=4000
+
+# Maximum loop count in the workflow
+NEXT_PUBLIC_LOOP_NODE_MAX_COUNT=100
+
+# Maximum number of tools in the agent/workflow
+NEXT_PUBLIC_MAX_TOOLS_NUM=10
+
+# Maximum number of Parallelism branches in the workflow
+NEXT_PUBLIC_MAX_PARALLEL_LIMIT=10
+
+# The maximum number of iterations for agent setting
+NEXT_PUBLIC_MAX_ITERATIONS_NUM=99
+
+NEXT_PUBLIC_ENABLE_WEBSITE_JINAREADER=true
+NEXT_PUBLIC_ENABLE_WEBSITE_FIRECRAWL=true
+NEXT_PUBLIC_ENABLE_WEBSITE_WATERCRAWL=true
+
+# The maximum number of tree node depth for workflow
+NEXT_PUBLIC_MAX_TREE_DEPTH=50