diff --git a/crates/df-ai-core/src/provider.rs b/crates/df-ai-core/src/provider.rs
index 26beef4..2d2bd33 100644
--- a/crates/df-ai-core/src/provider.rs
+++ b/crates/df-ai-core/src/provider.rs
@@ -36,6 +36,9 @@ pub struct CompletionRequest {
     /// 工具调用策略: "auto" | "none" | {"type":"function","name":"xxx"}
     #[serde(skip_serializing_if = "Option::is_none")]
     pub tool_choice: Option<serde_json::Value>,
+    /// DeepSeek thinking 模式：需把上一轮 response 的 reasoning_content 回传
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub reasoning_content: Option<String>,
 }
 
 /// 多模态消息内容片。Text 片为字符串；Image 片可走 url 或 base64（二选一，base64 非空时 url 忽略）。
@@ -115,29 +118,32 @@ pub struct ChatMessage {
     /// 默认 None（向前兼容老 JSON 反序列化）。落库随 messages JSON 序列化，无需独立列。
     #[serde(default, skip_serializing_if = "Option::is_none")]
     pub status: Option<String>,
+    /// DeepSeek thinking 模式的推理内容（多轮需回传）
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub reasoning_content: Option<String>,
 }
 
 impl ChatMessage {
     pub fn system(content: impl Into<String>) -> Self {
-        Self { role: MessageRole::System, content: content.into(), parts: None, tool_call_id: None, tool_calls: None, model: None, status: None }
+        Self { role: MessageRole::System, content: content.into(), parts: None, tool_call_id: None, tool_calls: None, model: None, status: None, reasoning_content: None }
     }
     pub fn user(content: impl Into<String>) -> Self {
-        Self { role: MessageRole::User, content: content.into(), parts: None, tool_call_id: None, tool_calls: None, model: None, status: None }
+        Self { role: MessageRole::User, content: content.into(), parts: None, tool_call_id: None, tool_calls: None, model: None, status: None, reasoning_content: None }
     }
     pub fn assistant(content: impl Into<String>) -> Self {
-        Self { role: MessageRole::Assistant, content: content.into(), parts: None, tool_call_id: None, tool_calls: None, model: None, status: None }
+        Self { role: MessageRole::Assistant, content: content.into(), parts: None, tool_call_id: None, tool_calls: None, model: None, status: None, reasoning_content: None }
     }
     pub fn assistant_with_tools(content: impl Into<String>, tool_calls: Vec<ToolCall>) -> Self {
-        Self { role: MessageRole::Assistant, content: content.into(), parts: None, tool_call_id: None, tool_calls: Some(tool_calls), model: None, status: None }
+        Self { role: MessageRole::Assistant, content: content.into(), parts: None, tool_call_id: None, tool_calls: Some(tool_calls), model: None, status: None, reasoning_content: None }
     }
     pub fn tool_result(call_id: impl Into<String>, content: impl Into<String>) -> Self {
-        Self { role: MessageRole::Tool, content: content.into(), parts: None, tool_call_id: Some(call_id.into()), tool_calls: None, model: None, status: None }
+        Self { role: MessageRole::Tool, content: content.into(), parts: None, tool_call_id: Some(call_id.into()), tool_calls: None, model: None, status: None, reasoning_content: None }
     }
 
     /// 多模态 user 消息：content 文本 + parts（含 Image 片）。
     /// content 作为人类可读文本（也作非 vision 端点降级载荷）；parts 透传给 vision 端点。
     pub fn user_parts(content: impl Into<String>, parts: Vec<ContentPart>) -> Self {
-        Self { role: MessageRole::User, content: content.into(), parts: Some(parts), tool_call_id: None, tool_calls: None, model: None, status: None }
+        Self { role: MessageRole::User, content: content.into(), parts: Some(parts), tool_call_id: None, tool_calls: None, model: None, status: None, reasoning_content: None }
     }
 
     /// 是否含图片片（供 provider 判定走多模态分支）。
@@ -249,6 +255,9 @@ pub struct CompletionResponse {
     /// AI 发起的工具调用（如有）
     #[serde(skip_serializing_if = "Option::is_none")]
     pub tool_calls: Option<Vec<ToolCall>>,
+    /// DeepSeek thinking 模式响应中的推理内容（需回传到下一轮请求）
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub reasoning_content: Option<String>,
 }
 
 /// Token 用量
@@ -276,6 +285,9 @@ pub struct StreamChunk {
     /// 非空表示流中途出错，不应视为正常完成（finished 路径），由 stream_llm 转 AiError。
     #[serde(skip)]
     pub error: Option<String>,
+    /// DeepSeek thinking 模式推理内容增量
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub reasoning_content: Option<String>,
 }
 
 /// 工具调用增量（流式中的片段）
@@ -449,8 +461,98 @@ mod tests {
             tool_calls: None,
             model: None,
             status: None,
+            reasoning_content: None,
         };
         assert_eq!(m.content, "字面量构造");
         assert!(m.parts.is_none());
     }
+
+    // ---------- DeepSeek reasoning_content ----------
+
+    /// 有 reasoning_content 时应序列化出来；None 时不出现
+    #[test]
+    fn completion_request_reasoning_content_serialization() {
+        let req = CompletionRequest {
+            model: "deepseek-r1".to_string(),
+            messages: vec![ChatMessage::user("hello")],
+            temperature: None,
+            max_tokens: None,
+            stream: false,
+            tools: None,
+            tool_choice: None,
+            reasoning_content: Some("let me think...".to_string()),
+        };
+        let json = serde_json::to_string(&req).unwrap();
+        assert!(json.contains("reasoning_content"), "reasoning_content 应出现在 JSON 中, got: {}", json);
+
+        // None 时不应出现（skip_serializing_if）
+        let req_none = CompletionRequest { reasoning_content: None, ..req };
+        let json_none = serde_json::to_string(&req_none).unwrap();
+        assert!(!json_none.contains("reasoning_content"), "None 的 reasoning_content 不应序列化, got: {}", json_none);
+    }
+
+    /// 所有便捷构造函数默认 reasoning_content 为 None
+    #[test]
+    fn chat_message_reasoning_content_constructors() {
+        assert!(ChatMessage::system("sys").reasoning_content.is_none());
+        assert!(ChatMessage::user("hi").reasoning_content.is_none());
+        assert!(ChatMessage::assistant("resp").reasoning_content.is_none());
+        assert!(ChatMessage::tool_result("call_1", "result").reasoning_content.is_none());
+    }
+
+    /// ChatMessage reasoning_content 序列化 round-trip
+    #[test]
+    fn chat_message_reasoning_content_roundtrip() {
+        let m = ChatMessage {
+            role: MessageRole::Assistant,
+            content: "answer".to_string(),
+            parts: None,
+            tool_call_id: None,
+            tool_calls: None,
+            model: None,
+            status: None,
+            reasoning_content: Some("thinking process".to_string()),
+        };
+        let json = serde_json::to_string(&m).unwrap();
+        assert!(json.contains("reasoning_content"));
+
+        let deserialized: ChatMessage = serde_json::from_str(&json).unwrap();
+        assert_eq!(deserialized.reasoning_content, Some("thinking process".to_string()));
+    }
+
+    /// StreamChunk reasoning_content 序列化
+    #[test]
+    fn stream_chunk_reasoning_content() {
+        let chunk = StreamChunk {
+            delta: "text".to_string(),
+            finished: false,
+            tool_calls: None,
+            usage: None,
+            error: None,
+            reasoning_content: Some("thought".to_string()),
+        };
+        let json = serde_json::to_string(&chunk).unwrap();
+        assert!(json.contains("reasoning_content"));
+
+        let chunk_none = StreamChunk { reasoning_content: None, ..chunk };
+        let json_none = serde_json::to_string(&chunk_none).unwrap();
+        assert!(!json_none.contains("reasoning_content"));
+    }
+
+    /// CompletionResponse reasoning_content 序列化
+    #[test]
+    fn completion_response_reasoning_content() {
+        let resp = CompletionResponse {
+            text: "ok".to_string(),
+            model: "r1".to_string(),
+            usage: TokenUsage { prompt_tokens: 10, completion_tokens: 20, total_tokens: 30 },
+            tool_calls: None,
+            reasoning_content: Some("r1 thought".to_string()),
+        };
+        let json = serde_json::to_string(&resp).unwrap();
+        assert!(json.contains("reasoning_content"));
+
+        let deserialized: CompletionResponse = serde_json::from_str(&json).unwrap();
+        assert_eq!(deserialized.reasoning_content, Some("r1 thought".to_string()));
+    }
 }
diff --git a/crates/df-ai/src/anthropic_compat.rs b/crates/df-ai/src/anthropic_compat.rs
index 90e91ea..02de838 100644
--- a/crates/df-ai/src/anthropic_compat.rs
+++ b/crates/df-ai/src/anthropic_compat.rs
@@ -109,7 +109,7 @@ pub(crate) fn apply_anthropic_event(data: &str, usage_accum: &mut Option<TokenUs
     let v: serde_json::Value = match serde_json::from_str(data) {
         Ok(v) => v,
         Err(_) => {
-            return StreamChunk { delta: String::new(), finished: false, tool_calls: None, usage: None, error: None }
+            return StreamChunk { delta: String::new(), finished: false, tool_calls: None, usage: None, error: None, reasoning_content: None }
         }
     };
     let ty = v.get("type").and_then(|t| t.as_str()).unwrap_or("");
@@ -128,7 +128,7 @@ pub(crate) fn apply_anthropic_event(data: &str, usage_accum: &mut Option<TokenUs
                     total_tokens: inp as u32,
                 });
             }
-            StreamChunk { delta: String::new(), finished: false, tool_calls: None, usage: None, error: None }
+            StreamChunk { delta: String::new(), finished: false, tool_calls: None, usage: None, error: None, reasoning_content: None }
         }
         // 消息增量：output_tokens 是累计值（非增量），直接覆盖 completion + 重算 total
         "message_delta" => {
@@ -138,14 +138,14 @@ pub(crate) fn apply_anthropic_event(data: &str, usage_accum: &mut Option<TokenUs
                 acc.completion_tokens = out as u32;
                 acc.total_tokens = acc.prompt_tokens + acc.completion_tokens;
             }
-            StreamChunk { delta: String::new(), finished: false, tool_calls: None, usage: None, error: None }
+            StreamChunk { delta: String::new(), finished: false, tool_calls: None, usage: None, error: None, reasoning_content: None }
         }
         // 文本增量
         "content_block_delta" => {
             if let Some(delta) = v.get("delta") {
                 if delta.get("type").and_then(|t| t.as_str()) == Some("text_delta") {
                     let text = delta.get("text").and_then(|t| t.as_str()).unwrap_or("").to_string();
-                    return StreamChunk { delta: text, finished: false, tool_calls: None, usage: None, error: None };
+                    return StreamChunk { delta: text, finished: false, tool_calls: None, usage: None, error: None, reasoning_content: None };
                 }
                 // 工具入参增量
                 if delta.get("type").and_then(|t| t.as_str()) == Some("input_json_delta") {
@@ -162,10 +162,11 @@ pub(crate) fn apply_anthropic_event(data: &str, usage_accum: &mut Option<TokenUs
                         }]),
                         usage: None,
                         error: None,
+                        reasoning_content: None,
                     };
                 }
             }
-            StreamChunk { delta: String::new(), finished: false, tool_calls: None, usage: None, error: None }
+            StreamChunk { delta: String::new(), finished: false, tool_calls: None, usage: None, error: None, reasoning_content: None }
         }
         // 工具块开始：带 id + name
         "content_block_start" => {
@@ -194,10 +195,11 @@ pub(crate) fn apply_anthropic_event(data: &str, usage_accum: &mut Option<TokenUs
                         }]),
                         usage: None,
                         error: None,
+                        reasoning_content: None,
                     };
                 }
             }
-            StreamChunk { delta: String::new(), finished: false, tool_calls: None, usage: None, error: None }
+            StreamChunk { delta: String::new(), finished: false, tool_calls: None, usage: None, error: None, reasoning_content: None }
         }
         // 消息结束：带出累积 usage
         "message_stop" => StreamChunk {
@@ -206,16 +208,17 @@ pub(crate) fn apply_anthropic_event(data: &str, usage_accum: &mut Option<TokenUs
             tool_calls: None,
             usage: usage_accum.take(),
             error: None,
+            reasoning_content: None,
         },
         // 错误事件：流中途出错。不走 finished 完成路径（避免残缺响应被当正常完成入库），
         // 改由 stream_llm 识别 error 非空 → 发 AiError + 丢弃残缺（与 OpenAI 路径 Err 一致）。
         "error" => {
             let msg = v.get("error").and_then(|e| e.get("message")).and_then(|m| m.as_str()).unwrap_or("stream error").to_string();
             error!(%msg, "Anthropic 流式错误事件");
-            StreamChunk { delta: String::new(), finished: false, tool_calls: None, usage: None, error: Some(msg) }
+            StreamChunk { delta: String::new(), finished: false, tool_calls: None, usage: None, error: Some(msg), reasoning_content: None }
         }
         // content_block_stop / ping 等不产出 chunk
-        _ => StreamChunk { delta: String::new(), finished: false, tool_calls: None, usage: None, error: None },
+        _ => StreamChunk { delta: String::new(), finished: false, tool_calls: None, usage: None, error: None, reasoning_content: None },
     }
 }
 
@@ -541,6 +544,7 @@ impl LlmProvider for AnthropicCompatProvider {
                     model: resp.model,
                     usage,
                     tool_calls: if tool_calls.is_empty() { None } else { Some(tool_calls) },
+                    reasoning_content: None,
                 })
             }
         })
@@ -808,6 +812,7 @@ mod tests {
             stream: false,
             tools: None,
             tool_choice: None,
+            reasoning_content: None,
         };
         let body = provider.convert_request(req);
         // user 消息 content 应为数组形态
@@ -839,6 +844,7 @@ mod tests {
             stream: false,
             tools: None,
             tool_choice: None,
+            reasoning_content: None,
         };
         let body = provider.convert_request(req);
         let user_msg = body
diff --git a/crates/df-ai/src/openai_compat.rs b/crates/df-ai/src/openai_compat.rs
index c94b8e7..cb73276 100644
--- a/crates/df-ai/src/openai_compat.rs
+++ b/crates/df-ai/src/openai_compat.rs
@@ -44,6 +44,9 @@ struct OpenAiRequest {
     /// 流式时请求末 chunk 携带 usage（OpenAI 官方 + DeepSeek/GLM 兼容）
     #[serde(skip_serializing_if = "Option::is_none")]
     stream_options: Option<serde_json::Value>,
+    /// DeepSeek thinking 模式
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub reasoning_content: Option<String>,
 }
 
 /// OpenAI 消息格式
@@ -58,6 +61,9 @@ struct OpenAiMessage {
     tool_call_id: Option<String>,
     #[serde(skip_serializing_if = "Option::is_none")]
     tool_calls: Option<Vec<serde_json::Value>>,
+    /// DeepSeek thinking 模式推理内容
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub reasoning_content: Option<String>,
 }
 
 /// OpenAI 同步响应
@@ -78,6 +84,9 @@ struct OpenAiChoice {
 struct OpenAiMessageResp {
     content: Option<String>,
     tool_calls: Option<Vec<OpenAiToolCallResp>>,
+    /// DeepSeek thinking 模式响应中的推理内容
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub reasoning_content: Option<String>,
 }
 
 #[derive(Debug, Deserialize)]
@@ -120,6 +129,9 @@ struct OpenAiStreamChoice {
 struct OpenAiStreamDelta {
     content: Option<String>,
     tool_calls: Option<Vec<OpenAiStreamToolCall>>,
+    /// DeepSeek thinking 模式流式推理内容
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub reasoning_content: Option<String>,
 }
 
 #[derive(Debug, Deserialize)]
@@ -158,6 +170,7 @@ pub(crate) fn apply_openai_sse(data: &str, usage_accum: &mut Option<TokenUsage>)
             tool_calls: None,
             usage: usage_accum.take(),
             error: None,
+            reasoning_content: None,
         };
     }
 
@@ -195,6 +208,7 @@ pub(crate) fn apply_openai_sse(data: &str, usage_accum: &mut Option<TokenUsage>)
                     tool_calls,
                     usage: None,
                     error: None,
+                    reasoning_content: choice.delta.reasoning_content,
                 }
             } else {
                 // choices 为空 = usage-only chunk，不输出文本（usage 已累积）
@@ -204,6 +218,7 @@ pub(crate) fn apply_openai_sse(data: &str, usage_accum: &mut Option<TokenUsage>)
                     tool_calls: None,
                     usage: None,
                     error: None,
+                    reasoning_content: None,
                 }
             }
         }
@@ -215,6 +230,7 @@ pub(crate) fn apply_openai_sse(data: &str, usage_accum: &mut Option<TokenUsage>)
                 tool_calls: None,
                 usage: None,
                 error: None,
+                reasoning_content: None,
             }
         }
     }
@@ -371,6 +387,7 @@ impl OpenAICompatProvider {
                     content,
                     tool_call_id: m.tool_call_id,
                     tool_calls,
+                    reasoning_content: m.reasoning_content,
                 }
             })
             .collect();
@@ -389,6 +406,7 @@ impl OpenAICompatProvider {
             stream: req.stream,
             tools,
             tool_choice: req.tool_choice,
+            reasoning_content: req.reasoning_content,
             // 流式请求末 chunk 带 usage（同步调用 complete 不需要）
             stream_options: if req.stream {
                 Some(serde_json::json!({ "include_usage": true }))
@@ -509,6 +527,7 @@ impl LlmProvider for OpenAICompatProvider {
                     model: body.model,
                     usage,
                     tool_calls,
+                    reasoning_content: choice.message.reasoning_content,
                 })
             }
         })
@@ -755,6 +774,7 @@ mod tests {
             stream: false,
             tools: None,
             tool_choice: None,
+            reasoning_content: None,
         };
         let out = provider.convert_request(req);
         let msg = &out.messages[0];
@@ -783,6 +803,7 @@ mod tests {
             stream: false,
             tools: None,
             tool_choice: None,
+            reasoning_content: None,
         };
         let out = provider.convert_request(req);
         let msg = &out.messages[0];
diff --git a/crates/df-nodes/src/ai_node.rs b/crates/df-nodes/src/ai_node.rs
index 618ce67..0d93fce 100644
--- a/crates/df-nodes/src/ai_node.rs
+++ b/crates/df-nodes/src/ai_node.rs
@@ -283,6 +283,7 @@ impl Node for AiNode {
             stream: false,
             tools: None,
             tool_choice: None,
+            reasoning_content: None,
         };
 
         tracing::info!(
@@ -513,6 +514,7 @@ impl Node for AiSelfReviewNode {
             stream: false,
             tools: None,
             tool_choice: None,
+            reasoning_content: None,
         };
 
         tracing::info!(
@@ -924,6 +926,7 @@ mod tests {
             stream: false,
             tools: None,
             tool_choice: None,
+            reasoning_content: None,
         };
         let response = provider.complete(request).await.expect("GLM 调用失败");
         assert!(!response.text.is_empty(), "GLM 返回空文本");
diff --git a/docs/todo.md b/docs/todo.md
index f1e3fd9..f8d22a8 100644
--- a/docs/todo.md
+++ b/docs/todo.md
@@ -115,6 +115,33 @@
 **已审文件清单（本轮）**：crates/df-ai-core/src/provider.rs · crates/df-ai/src/anthropic_compat.rs · crates/df-ai/src/openai_compat.rs · src-tauri/src/commands/ai/conversation.rs · src-tauri/src/commands/ai/commands.rs · src/api/ai.ts · src/api/types.ts · src/composables/ai/useAiSend.ts · src/composables/ai/useAiConversations.ts · src/stores/ai.ts · src/components/AiChat.vue（图片构建+渲染） · src/components/settings/ProviderPanel.vue
 **新登记 todo**：B-260617-07 · B-260617-08
 
+### 🔧 2026-06-17 Tasks.vue i18n 缺失（仅分析·未实施）
+
+> Tasks.vue:72 报 `Not found 'completed' key in 'zh' locale messages`。session-role-diagnose-only，仅分析+记todo。
+
+**现象**：Tasks.vue:72 `$t(statusLabel(task.status))` → `taskStatusLabel()` → `TASK_STATUS_LABELS[status]` 返回 `'tasks.status.done'` → `$t('tasks.status.done')` 在 en/zh-CN 均未定义 → intlify fallback 到 en 仍缺失。
+
+**根因**：`src/constants/project.ts:61-68` 定义了 `TASK_STATUS_LABELS`（值如 `'tasks.status.done'`/`'tasks.status.todo'` 等 7 个 status + `'tasks.statusFilter.done'` 等 5 个 filter + priority 相关），但 **en/zh-CN 的 tasks 相关 i18n 文件完全不存在**。6b67214 大量新增功能时未同步补 tasks i18n key。
+
+**影响范围**（需补全的 key）：
+| 组件 | 用到的 key 前缀 | 缺失的 key |
+|---|---|---|
+| Tasks.vue:72 | `tasks.status.*` (7个) | 全部 |
+| Tasks.vue:148-155 | `tasks.statusFilter.*` (5个) | 全部 |
+| Tasks.vue:96-103 | `tasks.priority.*` / `P0`/`P1`/`P2`/`P3` (5+4=9) | 全部 |
+
+- [ ] B-260617-10 [P2] — **Tasks.vue i18n 缺失**。`$t('tasks.status.done')` 等 **21 个翻译键**在 en/zh-CN 均未定义（intlify fallback 也找不到），根因：6b67214 新增大量功能时未同步补 tasks i18n key（`TASK_STATUS_LABELS` 定义了值但无对应 `$t()` 条目）。**影响**：Tasks.vue 状态标签/筛选器/优先级标签全部显示原始 key 而非翻译文本。—— src/views/Tasks.vue(:72) · src/constants/project.ts(:61-82 TASK_STATUS_LABELS/TASK_STATUS_CLASS/PRIORITY_LABELS/PRIORITY_CLASSES)
+
+> **2026-06-17 进展注记（1cd7652）**：提交 1cd7652 已做**数据层映射兜底**（project.ts 新增 `LEGACY_STATUS_MAP`：completed→done / review_ready→in_review / merged→done / abandoned→cancelled），`taskStatusLabel/taskStatusClass` 经映射后老 DB 脏态不再直接走 `$t()` 报 not found。**但 i18n key 补全（显示层）仍未做**——映射后 `$t('tasks.status.done')` 仍 not found（tasks.ts 不存在），用户从看到 "completed" 变成看到 "tasks.status.done" 原始 key。**B-260617-10 主问题（补全 21 个 i18n key）仍在**，1cd7652 是其前置的数据归一，待补 key 后老态自动正确显示。映射目标态（done/in_review/cancelled）均已核验在 TASK_STATUS_LABELS 7 态内 ✓。
+
+### 🔧 2026-06-17 走查·tauri 打包目标收窄 + 状态映射 DRY（仅分析·未实施）
+
+> 本轮 git diff 核验工作区未提交改动 + 最新提交 1cd7652。session-role-diagnose-only。
+
+- [ ] B-260617-11 [P2] — **tauri.conf.json 打包目标收窄未提交，若误入库锁死非 Windows 构建**。工作区改动 `bundle.targets: "all" → ["nsis"]`（src-tauri/tauri.conf.json:28），收窄到仅 Windows NSIS 安装包。若意图为本地只打 Windows 包，合理；**但若随其他改动一并提交**，macOS（dmg/app）、Linux（deb/appimage）构建将不可用，影响其他开发者/CI。**确认点**：该改动是临时本地构建还是有意入库？临时则建议提交前 revert 此行；有意则建议改为按平台条件配置而非硬编码单一 target。—— src-tauri/tauri.conf.json(:28)
+
+- [ ] B-260617-12 [P3·可选] — **LEGACY_STATUS_MAP 映射 DRY 重复**。1cd7652 中 `taskStatusLabel`(:96) 与 `taskStatusClass`(:101) 各写一遍 `const mapped = LEGACY_STATUS_MAP[status] ?? status`。两处一行重复，可抽 `mapLegacyStatus(status)` 收敛。极简优先可不动（仅 2 处 + 单行），登记备查。—— src/constants/project.ts(:96,:101)
+
 ### 🔧 2026-06-17 aichat 消息全量重叠（5角度深入分析·未实施）
 
 > 用户报 aichat 对话「大量重叠」「不单间距问题」。session-role-diagnose-only，5 角度并行论证后记录。
diff --git a/docs/待决策.md b/docs/待决策.md
index 56fdd15..e024f9b 100644
--- a/docs/待决策.md
+++ b/docs/待决策.md
@@ -79,6 +79,42 @@
 - **待用户操作**:部分场景运行时实测(A 路线场景 2/3)。
 - **状态**:🟡 待用户实测
 
+### 🟡 待设计/架构演进(有决策点,需方案设计或长期规划)
+
+#### ARC-260615-07 架构重构批(排期/优先级决策)
+- **背景**:src-tauri IPC 编排层 5711 行成事实业务层,7 项架构债。df-core→df-types 已完成(CR-61),剩 6 项独立大改。
+- **决策点**:6 项重构何时做/优先级/是否做(每项 ROI 与风险权衡,无法纯技术论证——做不做是资源/产品取舍)
+- **选项(剩余 6 项)**:
+  - a: df-app 抽取(IPC 编排层 5711 行独立 crate)
+  - b: AI agent loop 从 IPC 下沉 df-ai(逻辑与 IO 分离)
+  - c: 类型契约 ts-rs 代码生成(替代手写 types.ts)
+  - d: AiSession 多会话(B 路线前置,关联 S-260614-01)
+  - e: AppState 分组(5711 行 state 拆分)
+  - f: IPC 命名统一
+- **推荐**:**⏸️ 缓做**(重构低 ROI,当前功能优先;待稳定后按 d→b→a 顺序,d 是 B 路线前置可先)
+- **关联**:todo ARC-260615-07 / memory aichat-arch-extensibility
+- **状态**:🟡 待排期决策
+
+#### UX-260617-01 aichat 消息全量重叠(待 DevTools 定位根因)
+- **背景**:用户实测消息大量重叠堆叠。5 角度分析排除 CSS/scoped/DOM,核心嫌疑虚拟滚动 IO 半激活态不一致 + height=0 未设 minHeight → slot 塌 0 → 重叠。
+- **决策点**:根因确切触发路径(需 DevTools 实测,静态分析已到极限)
+- **选项**:
+  - a: 虚拟滚动 IO/RO 时序竞态(主嫌疑,unload 分支 minHeight 兜底修复)
+  - b: 其他(DOM 结构问题,需 DevTools 实查)
+- **推荐**:**待用户 DevTools 实测**(shouldRenderMsg 卸载时检查 slot 实际高度 + IO unobserve/observe 时序),定位后修法明确(unload 分支 minHeight fallback)
+- **关联**:todo UX-260617-01 / docs/05-代码审查/AI聊天组件极端数据场景分析-2026-06-17.md
+- **状态**:🟡 待 DevTools 验证根因
+
+#### UX-260617-28 双监听器同通道(长期架构演进)
+- **背景**:useAiEvents + useAiContext 各自 listen('ai-chat-event'),人工 AiCompressing flag 协调防双重处理,非架构保证。未来新增事件处理可能触发双重 bug。
+- **决策点**:是否升级单一分发器模式(架构保证)
+- **选项**:
+  - a: 保持现状(AiCompressing flag 协调够用)
+  - b: 单一分发器(架构保证,但大改)
+- **推荐**:**⏸️ 暂缓**(当前 flag 协调有效,未来新增事件处理触发双重 bug 时再升级)
+- **关联**:todo UX-260617-28 [INFO]
+- **状态**:⏸️ 长期(INFO,当前够用)
+
 ---
 
 ## 已决归档
diff --git a/src-tauri/src/commands/ai/agentic.rs b/src-tauri/src/commands/ai/agentic.rs
index b4ac4d7..67754ba 100644
--- a/src-tauri/src/commands/ai/agentic.rs
+++ b/src-tauri/src/commands/ai/agentic.rs
@@ -131,6 +131,8 @@ enum StreamOutcome {
         usage: df_ai::provider::TokenUsage,
         incomplete: bool,
         resolved_model: String,
+        /// DeepSeek thinking 模式推理内容（需回传到下一轮请求）
+        reasoning_content: Option<String>,
     },
     InitFailedExhausted,
     Fatal,
@@ -164,6 +166,7 @@ async fn stream_one_provider(
     retry_deadline: tokio::time::Instant,
     model_override: &Option<String>,
     agentic_req: &TaskRequirements,
+    last_reasoning_content: &Option<String>,
 ) -> StreamOutcome {
     // resolve→ensure→build 三步(复用 secret::build_provider_for,DRY)。
     // key 缺失/损坏 → Err → 归 Fatal(stream_llm Fatal 也是 key 类错,语义一致)。
@@ -209,17 +212,19 @@ async fn stream_one_provider(
             stream: true,
             tools: if tool_defs.is_empty() { None } else { Some(tool_defs.to_vec()) },
             tool_choice: None,
+            reasoning_content: last_reasoning_content.clone(),
         };
 
         match stream_llm(&*provider, retry_request, app_handle, stop_flag, notify, conv_id).await {
-            StreamResult::Complete { text, tool_calls, usage } => {
+            StreamResult::Complete { text, tool_calls, usage, reasoning_content } => {
                 return StreamOutcome::Success {
                     text, tool_calls, usage,
                     incomplete: false,
                     resolved_model,
+                    reasoning_content,
                 };
             }
-            StreamResult::Partial { text, tool_calls, usage } => {
+            StreamResult::Partial { text, tool_calls, usage, reasoning_content } => {
                 // MidStream 保文不重试(决策 a1)。携带 incomplete=true 交调用方走保文路径。
                 tracing::warn!(
                     conv_id = %conv_id,
@@ -231,6 +236,7 @@ async fn stream_one_provider(
                     text, tool_calls, usage,
                     incomplete: true,
                     resolved_model,
+                    reasoning_content,
                 };
             }
             StreamResult::InitFailed { retryable } => {
@@ -435,6 +441,9 @@ pub(crate) async fn run_agentic_loop(
     // 区分"正常收敛退出"与"达 MAX 被截断退出"——后者末轮 tool_calls 仍非空(tool_result 不再回传 LLM),属异常
     let mut converged = false;
 
+    // BUG-260617-12: DeepSeek thinking 模式推理内容跨轮透传
+    let mut last_reasoning_content: Option<String> = None;
+
     // F-260616-13: system_prompt 是 run_agentic_loop 的不变参数(整个 loop 期间文本不变),
     // 其 token 估算在 loop 外算一次缓存复用,避免每轮/每次重试重复 estimate_text(低收益优化,行为不变)。
     let sys_tokens = TokenEstimator::default().estimate_text(&system_prompt);
@@ -656,9 +665,9 @@ pub(crate) async fn run_agentic_loop(
         let mut candidate_chain: Vec<AiProviderRecord> = Vec::with_capacity(1 + candidates.len());
         candidate_chain.push(provider_config.clone());
         candidate_chain.extend(candidates.iter().cloned());
-        let (full_text, tool_calls_acc, round_usage, incomplete) = {
+        let (full_text, tool_calls_acc, round_usage, incomplete, round_reasoning_content) = {
             // outcome 累积成功结果(Complete/Partial),InitFailed 不写入。
-            let mut outcome: Option<(String, std::collections::HashMap<u32, super::ToolCallDraft>, df_ai::provider::TokenUsage, bool)> = None;
+            let mut outcome: Option<(String, std::collections::HashMap<u32, super::ToolCallDraft>, df_ai::provider::TokenUsage, bool, Option<String>)> = None;
 
             'candidate: for candidate in &candidate_chain {
                 // per-provider permit(可选):set_provider_caps 未配置时返回 None(单 provider 零变化);
@@ -678,14 +687,16 @@ pub(crate) async fn run_agentic_loop(
                     retry_deadline,
                     &model_override,
                     &agentic_req,
+                    &last_reasoning_content,
                 ).await {
-                    StreamOutcome::Success { text, tool_calls, usage, incomplete, resolved_model: m } => {
+                    StreamOutcome::Success { text, tool_calls, usage, incomplete, resolved_model: m, reasoning_content: rc } => {
                         // 成功:更新迭代级 resolved_model + provider_config(供后续 push/save/标题)。
                         // mut 解构(已在顶部声明 mut):本轮后续 save/push 用「实际成功所用 provider」
                         // 而非主 candidate。compress 已在本轮 stream 之前用过 primary,不受影响。
                         resolved_model = m;
                         provider_config = candidate.clone();
-                        outcome = Some((text, tool_calls, usage, incomplete));
+                        last_reasoning_content = rc;
+                        outcome = Some((text, tool_calls, usage, incomplete, last_reasoning_content.clone()));
                         break 'candidate;
                     }
                     StreamOutcome::InitFailedExhausted => {
@@ -750,6 +761,8 @@ pub(crate) async fn run_agentic_loop(
                 if !full_text.is_empty() {
                     let mut msg = ChatMessage::assistant(&full_text);
                     msg.model = Some(resolved_model.clone());
+                    // BUG-260617-12: MidStream 保文也回填 reasoning_content
+                    msg.reasoning_content = round_reasoning_content.clone();
                     session.messages.push(msg);
                     // 追加系统提示消息:响应因网络中断不完整(对齐决策 a1 系统提示机制)
                     let mut notice = ChatMessage::system("⚠ 响应因网络中断不完整,以上为已接收的部分内容。可重新发送以获取完整回复。");
@@ -807,10 +820,13 @@ pub(crate) async fn run_agentic_loop(
                     .collect();
                 let mut msg = ChatMessage::assistant_with_tools(&full_text, ai_tool_calls);
                 msg.model = Some(resolved_model.clone());
+                // BUG-260617-12: 回填 reasoning_content 供下一轮请求透传
+                msg.reasoning_content = last_reasoning_content.clone();
                 session.messages.push(msg);
             } else if !full_text.is_empty() {
                 let mut msg = ChatMessage::assistant(&full_text);
                 msg.model = Some(resolved_model.clone());
+                msg.reasoning_content = last_reasoning_content.clone();
                 session.messages.push(msg);
             }
         }
diff --git a/src-tauri/src/commands/ai/compress.rs b/src-tauri/src/commands/ai/compress.rs
index 67ba2fb..fa73c1b 100644
--- a/src-tauri/src/commands/ai/compress.rs
+++ b/src-tauri/src/commands/ai/compress.rs
@@ -77,6 +77,7 @@ pub(crate) async fn compress_via_llm(
         stream: false,
         tools: None,
         tool_choice: None,
+        reasoning_content: None,
     };
 
     // LLM 并发限流(压缩属独立调用,纳入双层 Semaphore)
diff --git a/src-tauri/src/commands/ai/knowledge_inject.rs b/src-tauri/src/commands/ai/knowledge_inject.rs
index 311fe72..b00a201 100644
--- a/src-tauri/src/commands/ai/knowledge_inject.rs
+++ b/src-tauri/src/commands/ai/knowledge_inject.rs
@@ -357,6 +357,7 @@ async fn extract_knowledge_from_conversation(
         stream: false,
         tools: None,
         tool_choice: None,
+        reasoning_content: None,
     };
 
     let provider: Box<dyn LlmProvider> = match super::secret::build_provider_for(provider_config) {
diff --git a/src-tauri/src/commands/ai/stream_recv.rs b/src-tauri/src/commands/ai/stream_recv.rs
index 91de8d1..5751e8f 100644
--- a/src-tauri/src/commands/ai/stream_recv.rs
+++ b/src-tauri/src/commands/ai/stream_recv.rs
@@ -101,6 +101,8 @@ pub(crate) enum StreamResult {
         text: String,
         tool_calls: HashMap<u32, ToolCallDraft>,
         usage: df_ai::provider::TokenUsage,
+        /// DeepSeek thinking 模式推理内容（需回传到下一轮请求）
+        reasoning_content: Option<String>,
     },
     /// 流中途断（MidStream chunk Err / idle timeout / provider stream-error / 有 partial_text
     /// 但流尽未收到 finished）。已 emit 过 AiTextDelta（前端 currentText 已累积），
@@ -110,6 +112,8 @@ pub(crate) enum StreamResult {
         text: String,
         tool_calls: HashMap<u32, ToolCallDraft>,
         usage: df_ai::provider::TokenUsage,
+        /// DeepSeek thinking 模式推理内容（部分累积）
+        reasoning_content: Option<String>,
     },
     /// Init 失败（provider.stream() Err：建连/鉴权/HTTP non-2xx）。已 emit AiError。
     /// `retryable`: 据状态码分类(retry::is_status_retryable 镜像)：true=5xx/429/timeout/connect
@@ -154,6 +158,8 @@ pub(crate) async fn stream_llm(
             let mut finished_received = false;
             let mut stopped = false;
             let mut final_usage: Option<df_ai::provider::TokenUsage> = None;
+            // BUG-260617-12: DeepSeek thinking 模式推理内容累积（多轮需回传）
+            let mut reasoning_content_acc: Option<String> = None;
 
             // B-260615-15:heartbeat interval 提至 loop 外复用,避免每轮重建计时器
             // (每轮重建会丢已积累的节拍,且 interval 首次 tick 立即返回的特性会被误用)。
@@ -208,6 +214,7 @@ pub(crate) async fn stream_llm(
                                     text: full_text,
                                     tool_calls: tool_calls_acc,
                                     usage: final_usage.unwrap_or_default(),
+                                    reasoning_content: reasoning_content_acc,
                                 };
                             }
                             Ok(None) => break, // 流正常结束
@@ -231,6 +238,10 @@ pub(crate) async fn stream_llm(
                                     if let Some(u) = &chunk.usage {
                                         final_usage = Some(u.clone());
                                     }
+                                    // BUG-260617-12: DeepSeek thinking 模式推理内容累积
+                                    if let Some(ref rc) = chunk.reasoning_content {
+                                        reasoning_content_acc.get_or_insert_with(String::new).push_str(rc);
+                                    }
                                     // provider 流式错误事件（Anthropic SSE `type=="error"` 等）：
                                     // UX-2025-04 / CR-30-2 / 决策 a1: MidStream 类失败——已有文本则保文(Partial),
                                     // 不重试。空文本无文可保,emit AiError + InitFailed(保守 retryable=true,
@@ -260,6 +271,7 @@ pub(crate) async fn stream_llm(
                                             text: full_text,
                                             tool_calls: tool_calls_acc,
                                             usage: final_usage.unwrap_or_default(),
+                                            reasoning_content: reasoning_content_acc,
                                         };
                                     }
                                     if chunk.finished {
@@ -301,6 +313,7 @@ pub(crate) async fn stream_llm(
                                         text: full_text,
                                         tool_calls: tool_calls_acc,
                                         usage: final_usage.unwrap_or_default(),
+                                        reasoning_content: reasoning_content_acc,
                                     };
                                 }
                             }
@@ -333,6 +346,7 @@ pub(crate) async fn stream_llm(
                     text: full_text,
                     tool_calls: tool_calls_acc,
                     usage: final_usage.unwrap_or_default(),
+                    reasoning_content: reasoning_content_acc,
                 };
             }
 
@@ -358,6 +372,7 @@ pub(crate) async fn stream_llm(
                     text: full_text,
                     tool_calls: tool_calls_acc,
                     usage: final_usage.unwrap_or_default(),
+                    reasoning_content: reasoning_content_acc,
                 };
             }
 
@@ -366,6 +381,7 @@ pub(crate) async fn stream_llm(
                 text: full_text,
                 tool_calls: tool_calls_acc,
                 usage: final_usage.unwrap_or_default(),
+                reasoning_content: reasoning_content_acc,
             }
         }
         Err(e) => {
diff --git a/src-tauri/src/commands/ai/title.rs b/src-tauri/src/commands/ai/title.rs
index 109c039..72fc86b 100644
--- a/src-tauri/src/commands/ai/title.rs
+++ b/src-tauri/src/commands/ai/title.rs
@@ -55,6 +55,7 @@ pub(crate) async fn ensure_conversation_title(
                 tool_calls: None,
                 model: None,
                 status: None,
+                reasoning_content: m.reasoning_content.clone(),
             })
             .collect();
         (summary, session.messages.all_messages_clone())
@@ -134,6 +135,7 @@ async fn generate_title_via_llm(
         stream: false,
         tools: None,
         tool_choice: None,
+        reasoning_content: None,
     };
     // LLM 并发限流（标题生成属独立调用，纳入双层 Semaphore）
     let _global_permit = llm_concurrency.acquire_global().await;