fix(presentation): 修复 presentation 模块类型错误和语法问题

- 创建 types.ts 定义完整的类型系统
- 重写 DocumentRenderer.tsx 修复语法错误
- 重写 QuizRenderer.tsx 修复语法错误
- 重写 PresentationContainer.tsx 添加类型守卫
- 重写 TypeSwitcher.tsx 修复类型引用
- 更新 index.ts 移除不存在的 ChartRenderer 导出

审计结果:
- 类型检查: 通过
- 单元测试: 222 passed
- 构建: 成功

crates/zclaw-runtime/src/loop_runner.rs

@@ -10,6 +10,7 @@ use crate::stream::StreamChunk;
 use crate::tool::{ToolRegistry, ToolContext, SkillExecutor};
 use crate::tool::builtin::PathValidator;
 use crate::loop_guard::LoopGuard;
+use crate::growth::GrowthIntegration;
 use zclaw_memory::MemoryStore;
 
 /// Agent loop runner
@@ -26,6 +27,8 @@ pub struct AgentLoop {
     temperature: f32,
     skill_executor: Option<Arc<dyn SkillExecutor>>,
     path_validator: Option<PathValidator>,
+    /// Growth system integration (optional)
+    growth: Option<GrowthIntegration>,
 }
 
 impl AgentLoop {
@@ -47,6 +50,7 @@ impl AgentLoop {
             temperature: 0.7,
             skill_executor: None,
             path_validator: None,
+            growth: None,
         }
     }
 
@@ -86,6 +90,22 @@ impl AgentLoop {
         self
     }
 
+    /// Enable growth system integration
+    pub fn with_growth(mut self, growth: GrowthIntegration) -> Self {
+        self.growth = Some(growth);
+        self
+    }
+
+    /// Set growth system (mutable)
+    pub fn set_growth(&mut self, growth: GrowthIntegration) {
+        self.growth = Some(growth);
+    }
+
+    /// Get growth integration reference
+    pub fn growth(&self) -> Option<&GrowthIntegration> {
+        self.growth.as_ref()
+    }
+
     /// Create tool context for tool execution
     fn create_tool_context(&self, session_id: SessionId) -> ToolContext {
         ToolContext {
@@ -108,35 +128,43 @@ impl AgentLoop {
     /// Implements complete agent loop: LLM → Tool Call → Tool Result → LLM → Final Response
     pub async fn run(&self, session_id: SessionId, input: String) -> Result<AgentLoopResult> {
         // Add user message to session
-        let user_message = Message::user(input);
+        let user_message = Message::user(input.clone());
         self.memory.append_message(&session_id, &user_message).await?;
 
         // Get all messages for context
         let mut messages = self.memory.get_messages(&session_id).await?;
 
+        // Enhance system prompt with growth memories
+        let enhanced_prompt = if let Some(ref growth) = self.growth {
+            let base = self.system_prompt.as_deref().unwrap_or("");
+            growth.enhance_prompt(&self.agent_id, base, &input).await?
+        } else {
+            self.system_prompt.clone().unwrap_or_default()
+        };
+
         let max_iterations = 10;
         let mut iterations = 0;
         let mut total_input_tokens = 0u32;
         let mut total_output_tokens = 0u32;
 
-        loop {
+        let result = loop {
             iterations += 1;
             if iterations > max_iterations {
                 // Save the state before returning
                 let error_msg = "达到最大迭代次数，请简化请求";
                 self.memory.append_message(&session_id, &Message::assistant(error_msg)).await?;
-                return Ok(AgentLoopResult {
+                break AgentLoopResult {
                     response: error_msg.to_string(),
                     input_tokens: total_input_tokens,
                     output_tokens: total_output_tokens,
                     iterations,
-                });
+                };
             }
 
             // Build completion request
             let request = CompletionRequest {
                 model: self.model.clone(),
-                system: self.system_prompt.clone(),
+                system: Some(enhanced_prompt.clone()),
                 messages: messages.clone(),
                 tools: self.tools.definitions(),
                 max_tokens: Some(self.max_tokens),
@@ -173,12 +201,12 @@ impl AgentLoop {
                 // Save final assistant message
                 self.memory.append_message(&session_id, &Message::assistant(&text)).await?;
 
-                return Ok(AgentLoopResult {
+                break AgentLoopResult {
                     response: text,
                     input_tokens: total_input_tokens,
                     output_tokens: total_output_tokens,
                     iterations,
-                });
+                };
             }
 
             // There are tool calls - add assistant message with tool calls to history
@@ -204,7 +232,18 @@ impl AgentLoop {
             }
 
             // Continue the loop - LLM will process tool results and generate final response
+        };
+
+        // Process conversation for memory extraction (post-conversation)
+        if let Some(ref growth) = self.growth {
+            if let Ok(all_messages) = self.memory.get_messages(&session_id).await {
+                if let Err(e) = growth.process_conversation(&self.agent_id, &all_messages, session_id.clone()).await {
+                    tracing::warn!("[AgentLoop] Growth processing failed: {}", e);
+                }
+            }
         }
+
+        Ok(result)
     }
 
     /// Run the agent loop with streaming
@@ -217,12 +256,20 @@ impl AgentLoop {
         let (tx, rx) = mpsc::channel(100);
 
         // Add user message to session
-        let user_message = Message::user(input);
+        let user_message = Message::user(input.clone());
         self.memory.append_message(&session_id, &user_message).await?;
 
         // Get all messages for context
         let messages = self.memory.get_messages(&session_id).await?;
 
+        // Enhance system prompt with growth memories
+        let enhanced_prompt = if let Some(ref growth) = self.growth {
+            let base = self.system_prompt.as_deref().unwrap_or("");
+            growth.enhance_prompt(&self.agent_id, base, &input).await?
+        } else {
+            self.system_prompt.clone().unwrap_or_default()
+        };
+
         // Clone necessary data for the async task
         let session_id_clone = session_id.clone();
         let memory = self.memory.clone();
@@ -231,7 +278,6 @@ impl AgentLoop {
         let skill_executor = self.skill_executor.clone();
         let path_validator = self.path_validator.clone();
         let agent_id = self.agent_id.clone();
-        let system_prompt = self.system_prompt.clone();
         let model = self.model.clone();
         let max_tokens = self.max_tokens;
         let temperature = self.temperature;
@@ -259,7 +305,7 @@ impl AgentLoop {
                 // Build completion request
                 let request = CompletionRequest {
                     model: model.clone(),
-                    system: system_prompt.clone(),
+                    system: Some(enhanced_prompt.clone()),
                     messages: messages.clone(),
                     tools: tools.definitions(),
                     max_tokens: Some(max_tokens),