diff --git a/src/chat/image_system/image_manager.py b/src/chat/image_system/image_manager.py
index 492886d4..c15f647f 100644
--- a/src/chat/image_system/image_manager.py
+++ b/src/chat/image_system/image_manager.py
@@ -44,6 +44,12 @@ class ImageManager:
 
         logger.info("图片管理器初始化完成")
 
+    def _get_image_record(self, image_hash: str) -> Optional[Images]:
+        """根据哈希获取图片记录。"""
+        with get_db_session() as session:
+            statement = select(Images).filter_by(image_hash=image_hash, image_type=ImageType.IMAGE).limit(1)
+            return session.exec(statement).first()
+
     async def get_image_description(
         self,
         *,
@@ -76,9 +82,8 @@ class ImageManager:
             hash_str = hashlib.sha256(image_bytes).hexdigest()
 
         try:
-            with get_db_session() as session:
-                statement = select(Images).filter_by(image_hash=hash_str, image_type=ImageType.IMAGE).limit(1)
-                if record := session.exec(statement).first():
+            if record := self._get_image_record(hash_str):
+                if record.vlm_processed and record.description:
                     return record.description
         except Exception as e:
             logger.error(f"查询图片描述时发生错误: {e}")
@@ -86,12 +91,17 @@ class ImageManager:
         if not image_bytes:
             logger.warning("图片哈希值未找到，且未提供图片字节数据，返回无描述")
             return ""
+        try:
+            await self.ensure_image_saved(image_bytes)
+        except Exception as e:
+            logger.error(f"保存图片文件时发生错误: {e}")
+            return ""
         if not wait_for_build:
             self._schedule_description_build(hash_str, image_bytes)
             return ""
         logger.info(f"图片描述未找到，哈希值: {hash_str}，准备生成新描述")
         try:
-            image = await self.save_image_and_process(image_bytes)
+            image = await self.build_image_description(image_bytes)
             return image.description
         except Exception as e:
             logger.error(f"生成图片描述时发生错误: {e}")
@@ -120,7 +130,7 @@ class ImageManager:
         """
         try:
             logger.info(f"图片描述后台构建已开始，哈希值: {image_hash}")
-            await self.save_image_and_process(image_bytes)
+            await self.build_image_description(image_bytes)
             logger.info(f"图片描述后台构建完成，哈希值: {image_hash}")
         except Exception as exc:
             logger.warning(f"图片描述后台构建失败，哈希值: {image_hash}，错误: {exc}")
@@ -201,6 +211,7 @@ class ImageManager:
                     return False
                 record.description = image.description
                 record.last_used_time = datetime.now()
+                record.vlm_processed = image.vlm_processed
                 session.add(record)
                 logger.info(f"成功更新图片描述: {image.file_hash}，新描述: {image.description}")
         except Exception as e:
@@ -239,22 +250,13 @@ class ImageManager:
             return False
         return True
 
-    async def save_image_and_process(self, image_bytes: bytes) -> MaiImage:
-        """
-        保存图片并生成描述
-
-        Args:
-            image_bytes (bytes): 图片的字节数据
-        Returns:
-            return (MaiImage): 包含图片信息的 MaiImage 对象
-        Raises:
-            Exception: 如果在保存或处理过程中发生错误
-        """
+    async def ensure_image_saved(self, image_bytes: bytes) -> MaiImage:
+        """先保存图片记录，确保后续可以按哈希回填图片内容。"""
         hash_str = hashlib.sha256(image_bytes).hexdigest()
 
         try:
             with get_db_session() as session:
-                statement = select(Images).filter_by(image_hash=hash_str).limit(1)
+                statement = select(Images).filter_by(image_hash=hash_str, image_type=ImageType.IMAGE).limit(1)
                 if record := session.exec(statement).first():
                     logger.info(f"图片已存在于数据库中，哈希值: {hash_str}")
                     record.last_used_time = datetime.now()
@@ -270,18 +272,38 @@ class ImageManager:
         tmp_file_path = IMAGE_DIR / f"{hash_str}.tmp"
         with tmp_file_path.open("wb") as f:
             f.write(image_bytes)
-        mai_image = MaiImage(full_path=(IMAGE_DIR / f"{hash_str}.tmp"), image_bytes=image_bytes)
+        mai_image = MaiImage(full_path=tmp_file_path, image_bytes=image_bytes)
         await mai_image.calculate_hash_format()
+        if not self.register_image_to_db(mai_image):
+            raise RuntimeError(f"保存图片记录到数据库失败: {hash_str}")
+        return mai_image
+
+    async def build_image_description(self, image_bytes: bytes) -> MaiImage:
+        """在图片已保存的前提下生成或补齐图片描述。"""
+        mai_image = await self.ensure_image_saved(image_bytes)
+        if mai_image.vlm_processed and mai_image.description:
+            return mai_image
+
         desc = await self._generate_image_description(image_bytes, mai_image.image_format)
         mai_image.description = desc
         mai_image.vlm_processed = True
-        try:
-            self.register_image_to_db(mai_image)
-        except Exception as e:
-            logger.error(f"保存新图片记录到数据库时发生错误: {e}")
-            raise e
+        if not self.update_image_description(mai_image):
+            raise RuntimeError(f"更新图片描述失败: {mai_image.file_hash}")
         return mai_image
 
+    async def save_image_and_process(self, image_bytes: bytes) -> MaiImage:
+        """
+        保存图片并生成描述
+
+        Args:
+            image_bytes (bytes): 图片的字节数据
+        Returns:
+            return (MaiImage): 包含图片信息的 MaiImage 对象
+        Raises:
+            Exception: 如果在保存或处理过程中发生错误
+        """
+        return await self.build_image_description(image_bytes)
+
     def cleanup_invalid_descriptions_in_db(self):
         """
         清理数据库中无效的图片记录
diff --git a/src/chat/message_receive/bot.py b/src/chat/message_receive/bot.py
index 27a13821..1015bf07 100644
--- a/src/chat/message_receive/bot.py
+++ b/src/chat/message_receive/bot.py
@@ -1,7 +1,6 @@
 """聊天消息入口与主链路调度。"""
 
 from contextlib import suppress
-from copy import deepcopy
 from typing import Any, Dict, List, Optional
 
 import os
@@ -550,9 +549,9 @@ class ChatBot:
             # if await self.handle_notice_message(message):
             #     pass
 
-            # 处理消息内容，识别表情包等二进制数据并转化为文本描述
-            if global_config.maisaka.direct_image_input:
-                message.maisaka_original_raw_message = deepcopy(message.raw_message)  # type: ignore[attr-defined]
+            # 处理消息内容，识别表情包等二进制数据并转化为文本描述。
+            # 如果 Maisaka 需要直接消费图片，会在后续构建 prompt 时按需回填图片二进制数据，
+            # 这里不再复制整条原始消息。
             # 入站主链优先保证消息尽快入队，避免图片、表情包、语音分析阻塞适配器超时。
             await message.process(
                 enable_heavy_media_analysis=False,
diff --git a/src/chat/message_receive/message.py b/src/chat/message_receive/message.py
index 3cf5fdf5..b2cb3aa6 100644
--- a/src/chat/message_receive/message.py
+++ b/src/chat/message_receive/message.py
@@ -1,11 +1,10 @@
+import asyncio
 from asyncio import Task
 from typing import Dict, List, Sequence, Tuple
 
 from rich.traceback import install
 from sqlmodel import select
 
-import asyncio
-
 from src.common.logger import get_logger
 from src.common.database.database import get_db_session
 from src.common.database.database_model import Messages
@@ -36,6 +35,102 @@ class MsgIDMapping:
 
 
 class SessionMessage(MaiMessage):
+    
+    #便于调试的打印函数
+    def __str__(self) -> str:
+        """返回适合日志输出的消息摘要。"""
+        return self.to_debug_string()
+
+    def __repr__(self) -> str:
+        """返回适合调试场景的消息摘要。"""
+        return self.to_debug_string()
+
+    def to_debug_string(self) -> str:
+        """构建包含引用信息的调试字符串。
+
+        Returns:
+            str: 适合记录日志的消息摘要。
+        """
+        user_info = self.message_info.user_info
+        group_info = self.message_info.group_info
+        chat_type = "group" if group_info else "private"
+        group_id = group_info.group_id if group_info else None
+        group_name = group_info.group_name if group_info else None
+        component_summaries = [self._summarize_component(component) for component in self.raw_message.components]
+        raw_components = ", ".join(component_summaries) if component_summaries else "empty"
+
+        return (
+            "SessionMessage("
+            f"message_id={self.message_id!r}, "
+            f"platform={self.platform!r}, "
+            f"chat_type={chat_type!r}, "
+            f"group_id={group_id!r}, "
+            f"group_name={group_name!r}, "
+            f"user_id={user_info.user_id!r}, "
+            f"user_nickname={user_info.user_nickname!r}, "
+            f"user_cardname={user_info.user_cardname!r}, "
+            f"reply_to={self.reply_to!r}, "
+            f"processed_plain_text={self._truncate_text(self.processed_plain_text)}, "
+            f"raw_components=[{raw_components}]"
+            ")"
+        )
+
+    @staticmethod
+    def _truncate_text(text: str | None, max_length: int = 120) -> str:
+        """截断较长文本，避免日志过长。
+
+        Args:
+            text: 原始文本。
+            max_length: 最大保留长度。
+
+        Returns:
+            str: 截断后的文本表示。
+        """
+        if text is None:
+            return "None"
+        normalized_text = text.replace("\r", "\\r").replace("\n", "\\n")
+        if len(normalized_text) <= max_length:
+            return repr(normalized_text)
+        return repr(f"{normalized_text[:max_length]}...")
+
+    def _summarize_component(self, component: StandardMessageComponents) -> str:
+        """生成单个消息组件的调试摘要。
+
+        Args:
+            component: 消息组件对象。
+
+        Returns:
+            str: 组件摘要文本。
+        """
+        if isinstance(component, TextComponent):
+            return f"Text(text={self._truncate_text(component.text, 80)})"
+        if isinstance(component, ImageComponent):
+            return f"Image(content={self._truncate_text(component.content or None, 60)})"
+        if isinstance(component, EmojiComponent):
+            return f"Emoji(content={self._truncate_text(component.content or None, 60)})"
+        if isinstance(component, AtComponent):
+            target_name = component.target_user_cardname or component.target_user_nickname or component.target_user_id
+            return f"At(target={target_name!r})"
+        if isinstance(component, VoiceComponent):
+            return f"Voice(content={self._truncate_text(component.content or None, 60)})"
+        if isinstance(component, ReplyComponent):
+            sender_name = (
+                component.target_message_sender_cardname
+                or component.target_message_sender_nickname
+                or component.target_message_sender_id
+            )
+            return (
+                "Reply("
+                f"target_message_id={component.target_message_id!r}, "
+                f"target_sender={sender_name!r}, "
+                f"target_content={self._truncate_text(component.target_message_content, 80)}"
+                ")"
+            )
+        if isinstance(component, ForwardNodeComponent):
+            return f"ForwardNode(count={len(component.forward_components)})"
+        return f"{component.__class__.__name__}"
+    #便于调试的打印函数end
+
     async def process(
         self,
         *,
diff --git a/src/chat/replyer/group_generator.py b/src/chat/replyer/group_generator.py
index 6858ca19..48b3a2cb 100644
--- a/src/chat/replyer/group_generator.py
+++ b/src/chat/replyer/group_generator.py
@@ -35,7 +35,7 @@ from src.services import llm_service as llm_api
 
 from src.chat.logger.plan_reply_logger import PlanReplyLogger
 from src.memory_system.memory_retrieval import init_memory_retrieval_sys, build_memory_retrieval_prompt
-from src.learners.jargon_explainer_old import explain_jargon_in_context, retrieve_concepts_with_jargon
+from src.learners.jargon_explainer_old import explain_jargon_in_context
 from src.chat.utils.common_utils import TempMethodsExpression
 
 init_memory_retrieval_sys()
diff --git a/src/chat/utils/statistic.py b/src/chat/utils/statistic.py
index 25add4bf..28982f22 100644
--- a/src/chat/utils/statistic.py
+++ b/src/chat/utils/statistic.py
@@ -15,7 +15,6 @@ from src.common.database.database import get_db_session
 from src.common.database.database_model import Messages, ModelUsage, OnlineTime, ToolRecord
 from src.manager.async_task_manager import AsyncTask
 from src.manager.local_store_manager import local_storage
-from src.config.config import global_config
 
 logger = get_logger("maibot_statistic")
 
diff --git a/src/maisaka/builtin_tool/context.py b/src/maisaka/builtin_tool/context.py
index 6bf3443d..4cf37986 100644
--- a/src/maisaka/builtin_tool/context.py
+++ b/src/maisaka/builtin_tool/context.py
@@ -6,7 +6,6 @@ from base64 import b64decode
 from datetime import datetime
 from typing import TYPE_CHECKING, Any, Dict, List, Optional
 
-from src.chat.message_receive.message import SessionMessage
 from src.chat.utils.utils import process_llm_response
 from src.common.data_models.message_component_data_model import EmojiComponent, MessageSequence, TextComponent
 from src.config.config import global_config
diff --git a/src/maisaka/chat_loop_service.py b/src/maisaka/chat_loop_service.py
index 8efb16bc..236277b6 100644
--- a/src/maisaka/chat_loop_service.py
+++ b/src/maisaka/chat_loop_service.py
@@ -15,7 +15,6 @@ from rich.panel import Panel
 
 from src.cli.console import console
 from src.common.data_models.llm_service_data_models import LLMGenerationOptions
-from src.common.data_models.message_component_data_model import MessageSequence, TextComponent
 from src.common.logger import get_logger
 from src.common.prompt_i18n import load_prompt
 from src.common.utils.utils_session import SessionUtils
@@ -38,9 +37,7 @@ from src.plugin_runtime.host.hook_spec_registry import HookSpec, HookSpecRegistr
 from src.services.llm_service import LLMServiceClient
 
 from .builtin_tool import get_builtin_tools
-from .context_messages import AssistantMessage, LLMContextMessage, SessionBackedMessage, ToolResultMessage
-from .message_adapter import format_speaker_content
-from .planner_message_utils import build_session_backed_text_message
+from .context_messages import AssistantMessage, LLMContextMessage, ToolResultMessage
 from .prompt_cli_renderer import PromptCLIVisualizer
 
 
@@ -324,7 +321,7 @@ class MaisakaChatLoopService:
         if not prompt_lines:
             return ""
 
-        return f"在该聊天中的注意事项：\n" + "\n\n".join(prompt_lines) + "\n"
+        return "在该聊天中的注意事项：\n" + "\n\n".join(prompt_lines) + "\n"
 
     @staticmethod
     def _get_chat_prompt_for_chat(chat_id: str, is_group_chat: Optional[bool]) -> str:
diff --git a/src/maisaka/context_messages.py b/src/maisaka/context_messages.py
index 174da097..65404d20 100644
--- a/src/maisaka/context_messages.py
+++ b/src/maisaka/context_messages.py
@@ -11,7 +11,13 @@ import base64
 from PIL import Image as PILImage
 
 from src.chat.message_receive.message import SessionMessage
-from src.common.data_models.message_component_data_model import EmojiComponent, ImageComponent, MessageSequence, TextComponent
+from src.common.data_models.message_component_data_model import (
+    EmojiComponent,
+    ImageComponent,
+    MessageSequence,
+    ReplyComponent,
+    TextComponent,
+)
 from src.llm_models.payload_content.message import Message, MessageBuilder, RoleType
 from src.llm_models.payload_content.tool_option import ToolCall
 
@@ -27,11 +33,42 @@ def _guess_image_format(image_bytes: bytes) -> Optional[str]:
         return None
 
 
-def _build_binary_component_type_text(component: EmojiComponent | ImageComponent) -> str:
-    """为图片类消息组件构造显式的消息类型标记。"""
-    if isinstance(component, EmojiComponent):
-        return "[消息类型]表情包"
-    return "[消息类型]图片"
+def _append_emoji_component(builder: MessageBuilder, component: EmojiComponent) -> bool:
+    """将表情组件追加到 LLM 消息构建器。"""
+    image_format = _guess_image_format(component.binary_data)
+    if image_format and component.binary_data:
+        builder.add_text_content("[消息类型]表情包")
+        builder.add_image_content(image_format, base64.b64encode(component.binary_data).decode("utf-8"))
+        return True
+
+    if component.content:
+        builder.add_text_content(component.content)
+        return True
+    return False
+
+
+def _append_image_component(builder: MessageBuilder, component: ImageComponent) -> bool:
+    """将图片组件追加到 LLM 消息构建器。"""
+    image_format = _guess_image_format(component.binary_data)
+    if image_format and component.binary_data:
+        builder.add_text_content("[消息类型]图片")
+        builder.add_image_content(image_format, base64.b64encode(component.binary_data).decode("utf-8"))
+        return True
+
+    if component.content:
+        builder.add_text_content(component.content)
+        return True
+    return False
+
+
+def _append_reply_component(builder: MessageBuilder, component: ReplyComponent) -> bool:
+    """将回复组件追加到 LLM 消息构建器。"""
+    target_message_id = component.target_message_id.strip()
+    if not target_message_id:
+        return False
+
+    builder.add_text_content(f"[引用回复]({target_message_id})")
+    return True
 
 
 def _build_message_from_sequence(
@@ -57,17 +94,17 @@ def _build_message_from_sequence(
                 has_content = True
             continue
 
-        if isinstance(component, (EmojiComponent, ImageComponent)):
-            image_format = _guess_image_format(component.binary_data)
-            if image_format and component.binary_data:
-                builder.add_text_content(_build_binary_component_type_text(component))
-                builder.add_image_content(image_format, base64.b64encode(component.binary_data).decode("utf-8"))
-                has_content = True
-                continue
+        if isinstance(component, EmojiComponent):
+            has_content = _append_emoji_component(builder, component) or has_content
+            continue
 
-            if component.content:
-                builder.add_text_content(component.content)
-                has_content = True
+        if isinstance(component, ImageComponent):
+            has_content = _append_image_component(builder, component) or has_content
+            continue
+
+        if isinstance(component, ReplyComponent):
+            has_content = _append_reply_component(builder, component) or has_content
+            continue
 
     if not has_content and fallback_text:
         builder.add_text_content(fallback_text)
diff --git a/src/maisaka/message_adapter.py b/src/maisaka/message_adapter.py
index b52d1baa..995f5b34 100644
--- a/src/maisaka/message_adapter.py
+++ b/src/maisaka/message_adapter.py
@@ -5,7 +5,13 @@ from datetime import datetime
 from typing import Optional
 import re
 
-from src.common.data_models.message_component_data_model import EmojiComponent, ImageComponent, MessageSequence, TextComponent
+from src.common.data_models.message_component_data_model import (
+    EmojiComponent,
+    ImageComponent,
+    MessageSequence,
+    ReplyComponent,
+    TextComponent,
+)
 
 SPEAKER_PREFIX_PATTERN = re.compile(
     r"^(?:(?P<timestamp>\d{2}:\d{2}:\d{2}))?(?:\[msg_id:(?P<message_id>[^\]]+)\])?\[(?P<speaker>[^\]]+)\](?P<content>.*)$",
@@ -65,5 +71,11 @@ def build_visible_text_from_sequence(message_sequence: MessageSequence) -> str:
 
         if isinstance(component, ImageComponent):
             parts.append("[图片]")
+            continue
+
+        if isinstance(component, ReplyComponent):
+            target_message_id = component.target_message_id.strip()
+            if target_message_id:
+                parts.append(f"[引用回复]({target_message_id})")
 
     return "".join(parts)
diff --git a/src/maisaka/reasoning_engine.py b/src/maisaka/reasoning_engine.py
index 1643f8c9..72539a5b 100644
--- a/src/maisaka/reasoning_engine.py
+++ b/src/maisaka/reasoning_engine.py
@@ -12,7 +12,7 @@ import traceback
 from src.chat.heart_flow.heartFC_utils import CycleDetail
 from src.chat.message_receive.message import SessionMessage
 from src.chat.utils.utils import process_llm_response
-from src.common.data_models.message_component_data_model import MessageSequence, TextComponent
+from src.common.data_models.message_component_data_model import EmojiComponent, ImageComponent, MessageSequence, TextComponent
 from src.common.logger import get_logger
 from src.config.config import global_config
 from src.core.tooling import ToolExecutionContext, ToolExecutionResult, ToolInvocation, ToolSpec
@@ -230,12 +230,10 @@ class MaisakaReasoningEngine:
         planner_prefix = build_planner_user_prefix_from_session_message(message)
 
         appended_component = False
-        if global_config.maisaka.direct_image_input:
-            source_sequence = getattr(message, "maisaka_original_raw_message", message.raw_message)
-        else:
-            source_sequence = message.raw_message
-
+        source_sequence = message.raw_message
         planner_components = clone_message_sequence(source_sequence).components
+        if global_config.maisaka.direct_image_input:
+            await self._hydrate_visual_components(planner_components)
         if planner_components and isinstance(planner_components[0], TextComponent):
             planner_components[0].text = planner_prefix + planner_components[0].text
         else:
@@ -256,6 +254,24 @@ class MaisakaReasoningEngine:
 
         return message_sequence, legacy_visible_text
 
+    async def _hydrate_visual_components(self, planner_components: list[object]) -> None:
+        """在 Maisaka 真正需要图片或表情时，按需回填二进制数据。"""
+        load_tasks: list[asyncio.Task[None]] = []
+        for component in planner_components:
+            if isinstance(component, ImageComponent) and not component.binary_data:
+                load_tasks.append(asyncio.create_task(component.load_image_binary()))
+                continue
+            if isinstance(component, EmojiComponent) and not component.binary_data:
+                load_tasks.append(asyncio.create_task(component.load_emoji_binary()))
+
+        if not load_tasks:
+            return
+
+        results = await asyncio.gather(*load_tasks, return_exceptions=True)
+        for result in results:
+            if isinstance(result, Exception):
+                logger.warning(f"{self._runtime.log_prefix} 回填图片或表情二进制数据失败，Maisaka 将退化为文本占位: {result}")
+
     def _build_legacy_visible_text(self, message: SessionMessage, source_sequence: MessageSequence) -> str:
         user_info = message.message_info.user_info
         speaker_name = user_info.user_cardname or user_info.user_nickname or user_info.user_id
diff --git a/src/mcp_module/connection.py b/src/mcp_module/connection.py
index c598e8bc..10a1ea5e 100644
--- a/src/mcp_module/connection.py
+++ b/src/mcp_module/connection.py
@@ -14,7 +14,7 @@ import httpx
 from src.cli.console import console
 from src.core.tooling import ToolExecutionResult
 
-from .config import MCPClientRuntimeConfig, MCPRootRuntimeConfig, MCPServerRuntimeConfig
+from .config import MCPClientRuntimeConfig, MCPServerRuntimeConfig
 from .hooks import MCPHostCallbacks
 from .models import (
     MCPPromptResult,