feat: add global memory mechanism for personalized conversations

Implement a memory system that stores user context and conversation history
in memory.json, uses LLM to summarize conversations, and injects relevant
context into system prompts for personalized responses.

Key components:
- MemoryConfig for configuration management
- MemoryUpdateQueue with debounce for batch processing
- MemoryUpdater for LLM-based memory extraction
- MemoryMiddleware to queue conversations after agent execution
- Memory injection into lead agent system prompt

Note: Add memory section to config.yaml to enable (see config.example.yaml)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

backend/src/agents/lead_agent/agent.py

@@ -4,6 +4,7 @@ from langchain_core.runnables import RunnableConfig
 
 from src.agents.lead_agent.prompt import apply_prompt_template
 from src.agents.middlewares.clarification_middleware import ClarificationMiddleware
+from src.agents.middlewares.memory_middleware import MemoryMiddleware
 from src.agents.middlewares.thread_data_middleware import ThreadDataMiddleware
 from src.agents.middlewares.title_middleware import TitleMiddleware
 from src.agents.middlewares.uploads_middleware import UploadsMiddleware
@@ -175,6 +176,8 @@ Being proactive with task management demonstrates thoroughness and ensures all r
 # UploadsMiddleware should be after ThreadDataMiddleware to access thread_id
 # SummarizationMiddleware should be early to reduce context before other processing
 # TodoListMiddleware should be before ClarificationMiddleware to allow todo management
+# TitleMiddleware generates title after first exchange
+# MemoryMiddleware queues conversation for memory update (after TitleMiddleware)
 # ViewImageMiddleware should be before ClarificationMiddleware to inject image details before LLM
 # ClarificationMiddleware should be last to intercept clarification requests after model calls
 def _build_middlewares(config: RunnableConfig):
@@ -202,6 +205,9 @@ def _build_middlewares(config: RunnableConfig):
     # Add TitleMiddleware
     middlewares.append(TitleMiddleware())
 
+    # Add MemoryMiddleware (after TitleMiddleware)
+    middlewares.append(MemoryMiddleware())
+
     # Add ViewImageMiddleware only if the current model supports vision
     model_name = config.get("configurable", {}).get("model_name") or config.get("configurable", {}).get("model")
     from src.config import get_app_config

backend/src/agents/lead_agent/prompt.py

@@ -7,6 +7,8 @@ SYSTEM_PROMPT_TEMPLATE = """
 You are DeerFlow 2.0, an open-source super agent.
 </role>
 
+{memory_context}
+
 <thinking_style>
 - Think concisely and strategically about the user's request BEFORE taking action
 - Break down the task: What is clear? What is ambiguous? What is missing?
@@ -164,6 +166,37 @@ The key AI trends for 2026 include enhanced reasoning capabilities, multimodal i
 """
 
 
+def _get_memory_context() -> str:
+    """Get memory context for injection into system prompt.
+
+    Returns:
+        Formatted memory context string wrapped in XML tags, or empty string if disabled.
+    """
+    try:
+        from src.agents.memory import format_memory_for_injection, get_memory_data
+        from src.config.memory_config import get_memory_config
+
+        config = get_memory_config()
+        if not config.enabled or not config.injection_enabled:
+            return ""
+
+        memory_data = get_memory_data()
+        memory_content = format_memory_for_injection(
+            memory_data, max_tokens=config.max_injection_tokens
+        )
+
+        if not memory_content.strip():
+            return ""
+
+        return f"""<memory>
+{memory_content}
+</memory>
+"""
+    except Exception as e:
+        print(f"Failed to load memory context: {e}")
+        return ""
+
+
 def apply_prompt_template() -> str:
     # Load only enabled skills
     skills = load_skills(enabled_only=True)
@@ -192,7 +225,14 @@ def apply_prompt_template() -> str:
     else:
         skills_list = "<!-- No skills available -->"
 
-    # Format the prompt with dynamic skills
-    prompt = SYSTEM_PROMPT_TEMPLATE.format(skills_list=skills_list, skills_base_path=container_base_path)
+    # Get memory context
+    memory_context = _get_memory_context()
+
+    # Format the prompt with dynamic skills and memory
+    prompt = SYSTEM_PROMPT_TEMPLATE.format(
+        skills_list=skills_list,
+        skills_base_path=container_base_path,
+        memory_context=memory_context,
+    )
 
     return prompt + f"\n<current_date>{datetime.now().strftime('%Y-%m-%d, %A')}</current_date>"