feat: Batch 5-9 — GrowthIntegration桥接、验证补全、死代码清理、Pipeline模板、Speech/Twitter真实实现

Batch 5 (P0): GrowthIntegration 接入 Tauri
- Kernel 新增 set_viking()/set_extraction_driver() 桥接 SqliteStorage
- 中间件链共享存储，MemoryExtractor 接入 LLM 驱动

Batch 6 (P1): 输入验证 + Heartbeat
- Relay 验证补全（stream 兼容检查、API key 格式校验）
- UUID 类型校验、SessionId 错误返回
- Heartbeat 默认开启 + 首次聊天自动初始化

Batch 7 (P2): 死代码清理
- zclaw-channels 整体移除（317 行）
- multi-agent 特性门控、admin 方法标注

Batch 8 (P2): Pipeline 模板
- PipelineMetadata 新增 annotations 字段
- pipeline_templates 命令 + 2 个示例模板
- fallback driver base_url 修复（doubao/qwen/deepseek 端点）

Batch 9 (P1): SpeechHand/TwitterHand 真实实现
- SpeechHand: tts_method 字段 + Browser TTS 前端集成 (Web Speech API)
- TwitterHand: 12 个 action 全部替换为 Twitter API v2 真实 HTTP 调用
- chatStore/useAutomationEvents 双路径 TTS 触发

desktop/src-tauri/src/intelligence/extraction_adapter.rs

@@ -246,6 +246,7 @@ pub fn is_extraction_driver_configured() -> bool {
 /// Get the global extraction driver.
 ///
 /// Returns `None` if not yet configured via `configure_extraction_driver`.
+#[allow(dead_code)]
 pub fn get_extraction_driver() -> Option<Arc<TauriExtractionDriver>> {
     EXTRACTION_DRIVER.get().cloned()
 }

desktop/src-tauri/src/intelligence/heartbeat.rs

@@ -100,12 +100,12 @@ pub type HeartbeatCheckFn = Box<dyn Fn(String) -> std::pin::Pin<Box<dyn std::fut
 impl Default for HeartbeatConfig {
     fn default() -> Self {
         Self {
-            enabled: false,
+            enabled: true,
             interval_minutes: 30,
             quiet_hours_start: Some("22:00".to_string()),
             quiet_hours_end: Some("08:00".to_string()),
             notify_channel: NotifyChannel::Ui,
-            proactivity_level: ProactivityLevel::Light,
+            proactivity_level: ProactivityLevel::Standard,
             max_alerts_per_tick: 5,
         }
     }

desktop/src-tauri/src/intelligence/validation.rs

@@ -57,6 +57,52 @@ impl fmt::Display for ValidationError {
 
 impl std::error::Error for ValidationError {}
 
+/// Validate a UUID string (for agent_id, session_id, etc.)
+///
+/// Provides a clear error message when the UUID format is invalid,
+/// instead of a generic "invalid characters" error from `validate_identifier`.
+pub fn validate_uuid(value: &str, field_name: &str) -> Result<(), ValidationError> {
+    let len = value.len();
+
+    if len == 0 {
+        return Err(ValidationError::RequiredFieldEmpty {
+            field: field_name.to_string(),
+        });
+    }
+
+    // UUID format: 8-4-4-4-12 hex digits with hyphens (36 chars total)
+    if len != 36 {
+        return Err(ValidationError::InvalidCharacters {
+            field: field_name.to_string(),
+            invalid_chars: format!("expected UUID format (36 chars), got {} chars", len),
+        });
+    }
+
+    // Quick structure check: positions 8,13,18,23 should be '-'
+    let bytes = value.as_bytes();
+    if bytes[8] != b'-' || bytes[13] != b'-' || bytes[18] != b'-' || bytes[23] != b'-' {
+        return Err(ValidationError::InvalidCharacters {
+            field: field_name.to_string(),
+            invalid_chars: "not a valid UUID (expected format: xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx)".into(),
+        });
+    }
+
+    // Check all non-hyphen positions are hex digits
+    for (i, &b) in bytes.iter().enumerate() {
+        if i == 8 || i == 13 || i == 18 || i == 23 {
+            continue;
+        }
+        if !b.is_ascii_hexdigit() {
+            return Err(ValidationError::InvalidCharacters {
+                field: field_name.to_string(),
+                invalid_chars: format!("'{}' at position {} is not a hex digit", b as char, i),
+            });
+        }
+    }
+
+    Ok(())
+}
+
 /// Validate an identifier (agent_id, pipeline_id, skill_id, etc.)
 ///
 /// # Rules

desktop/src-tauri/src/kernel_commands.rs

@@ -25,6 +25,11 @@ pub type SessionStreamGuard = Arc<dashmap::DashMap<String, Arc<Mutex<()>>>>;
 fn validate_agent_id(agent_id: &str) -> Result<String, String> {
     validate_identifier(agent_id, "agent_id")
         .map_err(|e| format!("Invalid agent_id: {}", e))?;
+    // AgentId is a UUID wrapper — validate UUID format for better error messages
+    if agent_id.contains('-') {
+        crate::intelligence::validation::validate_uuid(agent_id, "agent_id")
+            .map_err(|e| format!("Invalid agent_id: {}", e))?;
+    }
     Ok(agent_id.to_string())
 }
 
@@ -209,7 +214,7 @@ pub async fn kernel_init(
     let model = config.llm.model.clone();
 
     // Boot kernel
-    let kernel = Kernel::boot(config.clone())
+    let mut kernel = Kernel::boot(config.clone())
         .await
         .map_err(|e| format!("Failed to initialize kernel: {}", e))?;
 
@@ -222,6 +227,33 @@ pub async fn kernel_init(
         model.clone(),
     );
 
+    // Bridge SqliteStorage to Kernel's GrowthIntegration
+    // This connects the middleware chain (MemoryMiddleware, CompactionMiddleware)
+    // to the same persistent SqliteStorage used by viking_commands and intelligence_hooks.
+    {
+        match crate::viking_commands::get_storage().await {
+            Ok(sqlite_storage) => {
+                // Wrap SqliteStorage in VikingAdapter (SqliteStorage implements VikingStorage)
+                let viking = std::sync::Arc::new(zclaw_runtime::VikingAdapter::new(sqlite_storage));
+                kernel.set_viking(viking);
+                tracing::info!("[kernel_init] Bridged persistent SqliteStorage to Kernel GrowthIntegration");
+            }
+            Err(e) => {
+                tracing::warn!(
+                    "[kernel_init] Failed to get SqliteStorage, GrowthIntegration will use in-memory storage: {}",
+                    e
+                );
+            }
+        }
+
+        // Set the LLM extraction driver on the kernel for memory extraction via middleware
+        let extraction_driver = crate::intelligence::extraction_adapter::TauriExtractionDriver::new(
+            driver.clone(),
+            model.clone(),
+        );
+        kernel.set_extraction_driver(std::sync::Arc::new(extraction_driver));
+    }
+
     // Configure summary driver so the Growth system can generate L0/L1 summaries
     if let Some(api_key) = config_request.as_ref().and_then(|r| r.api_key.clone()) {
         crate::summarizer_adapter::configure_summary_driver(
@@ -530,6 +562,21 @@ pub async fn agent_chat_stream(
             format!("Session {} already has an active stream", session_id)
         })?;
 
+    // AUTO-INIT HEARTBEAT: Ensure heartbeat engine exists for this agent.
+    // Uses default config (enabled: true, 30min interval) so heartbeat runs
+    // automatically from the first conversation without manual setup.
+    {
+        let mut engines = heartbeat_state.lock().await;
+        if !engines.contains_key(&request.agent_id) {
+            let engine = crate::intelligence::heartbeat::HeartbeatEngine::new(
+                request.agent_id.clone(),
+                None, // Use default config (enabled: true)
+            );
+            engines.insert(request.agent_id.clone(), engine);
+            tracing::info!("[agent_chat_stream] Auto-initialized heartbeat for agent: {}", request.agent_id);
+        }
+    }
+
     // PRE-CONVERSATION: Build intelligence-enhanced system prompt
     let enhanced_prompt = crate::intelligence_hooks::pre_conversation_hook(
         &request.agent_id,
@@ -550,15 +597,22 @@ pub async fn agent_chat_stream(
         // Use intelligence-enhanced system prompt if available
         let prompt_arg = if enhanced_prompt.is_empty() { None } else { Some(enhanced_prompt) };
         // Parse session_id for session reuse (carry conversation history across turns)
-        let session_id_parsed = std::str::FromStr::from_str(&session_id)
-            .ok()
-            .map(|uuid| zclaw_types::SessionId::from_uuid(uuid));
-        if session_id_parsed.is_none() {
-            tracing::warn!(
-                "session_id '{}' is not a valid UUID, will create a new session (context will be lost)",
-                session_id
-            );
-        }
+        // Empty session_id means first message in a new conversation — that's valid.
+        // Non-empty session_id MUST be a valid UUID; if not, return error instead of
+        // silently losing context by creating a new session.
+        let session_id_parsed = if session_id.is_empty() {
+            None
+        } else {
+            match uuid::Uuid::parse_str(&session_id) {
+                Ok(uuid) => Some(zclaw_types::SessionId::from_uuid(uuid)),
+                Err(e) => {
+                    return Err(format!(
+                        "Invalid session_id '{}': {}. Cannot reuse conversation context.",
+                        session_id, e
+                    ));
+                }
+            }
+        };
         let rx = kernel.send_message_stream_with_prompt(&id, message.clone(), prompt_arg, session_id_parsed)
             .await
             .map_err(|e| format!("Failed to start streaming: {}", e))?;
@@ -1775,9 +1829,10 @@ pub async fn scheduled_task_list(
 }
 
 // ============================================================
-// A2A (Agent-to-Agent) Commands
+// A2A (Agent-to-Agent) Commands — gated behind multi-agent feature
 // ============================================================
 
+#[cfg(feature = "multi-agent")]
 /// Send a direct A2A message from one agent to another
 #[tauri::command]
 pub async fn agent_a2a_send(
@@ -1810,6 +1865,7 @@ pub async fn agent_a2a_send(
 }
 
 /// Broadcast a message from one agent to all other agents
+#[cfg(feature = "multi-agent")]
 #[tauri::command]
 pub async fn agent_a2a_broadcast(
     state: State<'_, KernelState>,
@@ -1830,6 +1886,7 @@ pub async fn agent_a2a_broadcast(
 }
 
 /// Discover agents with a specific capability
+#[cfg(feature = "multi-agent")]
 #[tauri::command]
 pub async fn agent_a2a_discover(
     state: State<'_, KernelState>,
@@ -1850,6 +1907,7 @@ pub async fn agent_a2a_discover(
   }
 
 /// Delegate a task to another agent and wait for response
+#[cfg(feature = "multi-agent")]
 #[tauri::command]
 pub async fn agent_a2a_delegate_task(
     state: State<'_, KernelState>,

desktop/src-tauri/src/lib.rs

@@ -1352,15 +1352,19 @@ pub fn run() {
             kernel_commands::scheduled_task_create,
             kernel_commands::scheduled_task_list,
 
-            // A2A commands (Agent-to-Agent messaging)
+            // A2A commands gated behind multi-agent feature
+            #[cfg(feature = "multi-agent")]
             kernel_commands::agent_a2a_send,
+            #[cfg(feature = "multi-agent")]
             kernel_commands::agent_a2a_broadcast,
+            #[cfg(feature = "multi-agent")]
             kernel_commands::agent_a2a_discover,
+            #[cfg(feature = "multi-agent")]
             kernel_commands::agent_a2a_delegate_task,
 
             // Pipeline commands (DSL-based workflows)
             pipeline_commands::pipeline_list,
-            pipeline_commands::pipeline_get,
+            pipeline_commands::pipeline_templates,            pipeline_commands::pipeline_get,
             pipeline_commands::pipeline_run,
             pipeline_commands::pipeline_progress,
             pipeline_commands::pipeline_cancel,

desktop/src-tauri/src/pipeline_commands.rs

@@ -681,9 +681,8 @@ fn scan_pipelines_with_paths(
                 tracing::debug!("[scan] File content length: {} bytes", content.len());
                 match parse_pipeline_yaml(&content) {
                     Ok(pipeline) => {
-                        // Debug: log parsed pipeline metadata
-                        println!(
-                            "[DEBUG scan] Parsed YAML: {} -> category: {:?}, industry: {:?}",
+                        tracing::debug!(
+                            "[scan] Parsed YAML: {} -> category: {:?}, industry: {:?}",
                             pipeline.metadata.name,
                             pipeline.metadata.category,
                             pipeline.metadata.industry
@@ -744,8 +743,8 @@ fn scan_pipelines_full_sync(
 
 fn pipeline_to_info(pipeline: &Pipeline) -> PipelineInfo {
     let industry = pipeline.metadata.industry.clone().unwrap_or_default();
-    println!(
-        "[DEBUG pipeline_to_info] Pipeline: {}, category: {:?}, industry: {:?}",
+    tracing::debug!(
+        "[pipeline_to_info] Pipeline: {}, category: {:?}, industry: {:?}",
         pipeline.metadata.name,
         pipeline.metadata.category,
         pipeline.metadata.industry
@@ -1040,16 +1039,30 @@ fn create_llm_driver_from_config() -> Option<Arc<dyn LlmActionDriver>> {
     // Convert api_key to SecretString
     let secret_key = SecretString::new(api_key);
 
-    // Create the runtime driver
+    // Create the runtime driver — use with_base_url when a custom endpoint is configured
+    // (essential for Chinese providers like doubao, qwen, deepseek, kimi)
     let runtime_driver: Arc<dyn zclaw_runtime::LlmDriver> = match provider.as_str() {
         "anthropic" => {
-            Arc::new(zclaw_runtime::AnthropicDriver::new(secret_key))
+            if let Some(url) = base_url {
+                Arc::new(zclaw_runtime::AnthropicDriver::with_base_url(secret_key, url))
+            } else {
+                Arc::new(zclaw_runtime::AnthropicDriver::new(secret_key))
+            }
         }
-        "openai" | "doubao" | "qwen" | "deepseek" | "kimi" => {
-            Arc::new(zclaw_runtime::OpenAiDriver::new(secret_key))
+        "openai" | "doubao" | "qwen" | "deepseek" | "kimi" | "zhipu" => {
+            // Chinese providers typically need a custom base_url
+            if let Some(url) = base_url {
+                Arc::new(zclaw_runtime::OpenAiDriver::with_base_url(secret_key, url))
+            } else {
+                Arc::new(zclaw_runtime::OpenAiDriver::new(secret_key))
+            }
         }
         "gemini" => {
-            Arc::new(zclaw_runtime::GeminiDriver::new(secret_key))
+            if let Some(url) = base_url {
+                Arc::new(zclaw_runtime::GeminiDriver::with_base_url(secret_key, url))
+            } else {
+                Arc::new(zclaw_runtime::GeminiDriver::new(secret_key))
+            }
         }
         "local" | "ollama" => {
             let url = base_url.unwrap_or_else(|| "http://localhost:11434".to_string());
@@ -1077,3 +1090,83 @@ pub async fn analyze_presentation(
     // Convert analysis to JSON
     serde_json::to_value(&analysis).map_err(|e| e.to_string())
 }
+
+/// Pipeline template metadata
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(rename_all = "camelCase")]
+pub struct PipelineTemplateInfo {
+    pub id: String,
+    pub display_name: String,
+    pub description: String,
+    pub category: String,
+    pub industry: String,
+    pub tags: Vec<String>,
+    pub icon: String,
+    pub version: String,
+    pub author: String,
+    pub inputs: Vec<PipelineInputInfo>,
+}
+
+/// List available pipeline templates from the `_templates/` directory.
+///
+/// Templates are pipeline YAML files that users can browse and instantiate.
+/// They live in `pipelines/_templates/` and are not directly runnable
+/// (they serve as blueprints).
+#[tauri::command]
+pub async fn pipeline_templates(
+    state: State<'_, Arc<PipelineState>>,
+) -> Result<Vec<PipelineTemplateInfo>, String> {
+    let pipelines = state.pipelines.read().await;
+
+    // Filter pipelines that have `is_template: true` in metadata
+    // or are in the _templates directory
+    let templates: Vec<PipelineTemplateInfo> = pipelines.iter()
+        .filter_map(|(id, pipeline)| {
+            // Check if this pipeline has template metadata
+            let is_template = pipeline.metadata.annotations
+                .as_ref()
+                .and_then(|a| a.get("is_template"))
+                .and_then(|v| v.as_bool())
+                .unwrap_or(false);
+
+            if !is_template {
+                return None;
+            }
+
+            Some(PipelineTemplateInfo {
+                id: pipeline.metadata.name.clone(),
+                display_name: pipeline.metadata.display_name.clone()
+                    .unwrap_or_else(|| pipeline.metadata.name.clone()),
+                description: pipeline.metadata.description.clone().unwrap_or_default(),
+                category: pipeline.metadata.category.clone().unwrap_or_default(),
+                industry: pipeline.metadata.industry.clone().unwrap_or_default(),
+                tags: pipeline.metadata.tags.clone(),
+                icon: pipeline.metadata.icon.clone().unwrap_or_else(|| "📦".to_string()),
+                version: pipeline.metadata.version.clone(),
+                author: pipeline.metadata.author.clone().unwrap_or_default(),
+                inputs: pipeline.spec.inputs.iter().map(|input| {
+                    PipelineInputInfo {
+                        name: input.name.clone(),
+                        input_type: match input.input_type {
+                            zclaw_pipeline::InputType::String => "string".to_string(),
+                            zclaw_pipeline::InputType::Number => "number".to_string(),
+                            zclaw_pipeline::InputType::Boolean => "boolean".to_string(),
+                            zclaw_pipeline::InputType::Select => "select".to_string(),
+                            zclaw_pipeline::InputType::MultiSelect => "multi-select".to_string(),
+                            zclaw_pipeline::InputType::File => "file".to_string(),
+                            zclaw_pipeline::InputType::Text => "text".to_string(),
+                        },
+                        required: input.required,
+                        label: input.label.clone().unwrap_or_else(|| input.name.clone()),
+                        placeholder: input.placeholder.clone(),
+                        default: input.default.clone(),
+                        options: input.options.clone(),
+                    }
+                }).collect(),
+            })
+        })
+        .collect();
+
+    tracing::debug!("[pipeline_templates] Found {} templates", templates.len());
+    Ok(templates)
+}