diff --git a/amrita_plugin_memory/cli.py b/amrita_plugin_memory/cli.py
index 012b5e7..021ec90 100644
--- a/amrita_plugin_memory/cli.py
+++ b/amrita_plugin_memory/cli.py
@@ -59,9 +59,6 @@ def _progress(done: int, total: int) -> None:
     return _progress
 
 
-#  status
-
-
 def cmd_status() -> None:
     """显示嵌入指纹、记忆分布与备份列表。"""
     collection = _get_collection()
@@ -118,9 +115,6 @@ def _print_scope_counts(collection: Collection) -> None:
         click.echo(f"  {scope_id}: {count}")
 
 
-#  reindex
-
-
 def cmd_reindex(*, assume_yes: bool = False) -> None:
     """用当前嵌入模型全量重映射。"""
     collection = _get_collection()
@@ -138,9 +132,6 @@ def cmd_reindex(*, assume_yes: bool = False) -> None:
     click.echo(f"完成，已重新嵌入 {done} 条记忆")
 
 
-#  migrate-keys
-
-
 def cmd_migrate_keys(*, dry_run: bool = False) -> None:
     """把分区键迁移到当前 Amrita 的 uni_id 格式。"""
     client = get_db_conn()
@@ -156,9 +147,6 @@ def cmd_migrate_keys(*, dry_run: bool = False) -> None:
         click.echo(f"完成，已改写 {changed} 条记忆的分区键")
 
 
-#  reset-fingerprint
-
-
 def cmd_reset_fingerprint() -> None:
     """只写入当前指纹、不重嵌入（逃生舱）。"""
     collection = _get_collection(create=True)
@@ -169,9 +157,6 @@ def cmd_reset_fingerprint() -> None:
     click.echo("⚠️  未重新嵌入 —— 若模型确实已变更，检索结果将不准确")
 
 
-#  backup
-
-
 def cmd_backup_list() -> None:
     """列出备份。"""
     backups = list_backups()
@@ -219,9 +204,6 @@ def cmd_backup_restore(file: str) -> None:
     click.echo(f"完成，已恢复 {done} 条记忆")
 
 
-#  click 命令组
-
-
 @click.group()
 def memory() -> None:
     """amrita_plugin_memory 维护命令"""
diff --git a/amrita_plugin_memory/config.py b/amrita_plugin_memory/config.py
index 03049b1..96c63ee 100644
--- a/amrita_plugin_memory/config.py
+++ b/amrita_plugin_memory/config.py
@@ -1,4 +1,3 @@
-# Configuration for your_plugin_name plugin
 from typing import Literal
 
 from amrita_core import ModelPreset
@@ -14,11 +13,15 @@
 
 
 class SubconsciousConfig(BaseModel):
-    """常驻推理循环（潜意识层）配置 — 实验性功能"""
+    """常驻推理循环（潜意识层）配置。
+
+    默认只实现单用户：多用户场景与用户体量难以预测，需要多用户支持时请自行实现。
+    """
 
     enabled: bool = Field(default=False, description="是否启用常驻推理循环")
     target_user_id: str = Field(
-        default="", description="目标用户ID（MVP仅支持单用户），为空则不启动"
+        default="",
+        description=("目标用户ID（默认仅实现单用户，多用户需自行实现），为空则不启动"),
     )
     allowed_tools: list[str] = Field(
         default_factory=list,
diff --git a/amrita_plugin_memory/embedding.py b/amrita_plugin_memory/embedding.py
index 5b84f86..bd99286 100644
--- a/amrita_plugin_memory/embedding.py
+++ b/amrita_plugin_memory/embedding.py
@@ -99,9 +99,6 @@ def describe_stored(collection: Collection) -> str:
     return f"{protocol} / {model} @ {base_url}"
 
 
-#  备份
-
-
 def _prune_backups(keep: int) -> None:
     backups = sorted(BACKUP_DIR.glob("embed_backup_*.json"))
     for stale in backups[:-keep] if keep > 0 else backups:
@@ -131,9 +128,6 @@ def list_backups() -> list[Path]:
     return sorted(BACKUP_DIR.glob("embed_backup_*.json"))
 
 
-#  全量重映射
-
-
 async def _write_in_batches(
     collection: Collection,
     ids: list[str],
@@ -265,9 +259,6 @@ def _run_async(coro: Any) -> Any:
     )
 
 
-#  启动检查
-
-
 def _confirm_reindex(reason: str, stored_desc: str) -> bool:
     """交互确认；非 TTY 时抛出 _RefuseToLoad。"""
     if not sys.stdin.isatty():
@@ -322,7 +313,6 @@ def _check_fingerprint(collection: Collection) -> None:
         _do_reindex_with_progress()
         return
 
-    # policy == "ask"
     stored_desc = describe_stored(collection) if stored else "（无记录）"
     if not _confirm_reindex(reason, stored_desc):
         logger.warning("[Memory] 用户拒绝重映射，保持现有数据不变。")
diff --git a/amrita_plugin_memory/keys.py b/amrita_plugin_memory/keys.py
index d9af435..25beb76 100644
--- a/amrita_plugin_memory/keys.py
+++ b/amrita_plugin_memory/keys.py
@@ -1,12 +1,12 @@
 """L2 向量层分区键 — 统一跟随已安装 Amrita 的 uni_id 格式。
 
-Amrita 的会话 ID 格式随版本演进：
+Amrita 的会话 ID 存在以下两种格式，本模块需要同时识别：
 
-- 旧版（≤1.9.x）：``user_{qq}`` / ``group_{群号}``
-- 新版（开发中）：``QQPlatform_Private_{qq}`` / ``QQPlatform_Group_{群号}``
+- ``user_{qq}`` / ``group_{群号}``
+- ``QQPlatform_Private_{qq}`` / ``QQPlatform_Group_{群号}``
 
 本模块**不硬编码任一格式**，而是委托框架的 ``make_uni_id`` 生成，
-并提供一个能识别两种历史格式的解析器，用于存量数据的 Key 迁移。
+并提供一个能识别两种格式的解析器，用于存量数据的 Key 迁移。
 """
 
 from __future__ import annotations
@@ -22,14 +22,14 @@
 
 Scope = Literal["group", "user"]
 
-#  集合 metadata 中的 Key 体系版本号，用于幂等迁移
+# 集合 metadata 中的 Key 体系版本号，用于幂等迁移
 KEY_SCHEMA_VERSION = 2
 KEY_SCHEMA_VERSION_META = "key_schema_version"
 
 _GROUP_KINDS = {"group", "Group"}
 _PRIVATE_KINDS = {"user", "Private"}
 
-#  同时匹配新旧两种格式（可选平台前缀 + 类型 + 数字 payload）
+# 同时匹配两种格式（可选平台前缀 + 类型 + 数字 payload）
 _ANY_ID_RE = re.compile(
     r"^(?:[A-Za-z0-9]+_)?(?P<kind>Private|Group|Channel|user|group)_(?P<payload>[0-9]+)$"
 )
diff --git a/amrita_plugin_memory/matchers.py b/amrita_plugin_memory/matchers.py
index 8bb2fad..c38f49d 100644
--- a/amrita_plugin_memory/matchers.py
+++ b/amrita_plugin_memory/matchers.py
@@ -184,11 +184,9 @@ async def _handle_delete(
         result = await ope.get_all_notes(partition_id, include=["metadatas"])
         all_ids: list[str] = result.get("ids") or []
 
-        # 精确匹配
         if doc_id in all_ids:
             resolved_id = doc_id
         else:
-            # 前缀匹配
             matches = [mid for mid in all_ids if mid.startswith(doc_id)]
             if len(matches) == 0:
                 await matcher.finish(
diff --git a/amrita_plugin_memory/migrations/21f55abc2b90_init.py b/amrita_plugin_memory/migrations/21f55abc2b90_init.py
index 2a65e43..662a67a 100644
--- a/amrita_plugin_memory/migrations/21f55abc2b90_init.py
+++ b/amrita_plugin_memory/migrations/21f55abc2b90_init.py
@@ -22,7 +22,6 @@
 def upgrade(name: str = "") -> None:
     if name:
         return
-    # ### commands auto generated by Alembic - please adjust! ###
     op.create_table('amrita_plugin_memory_user_memo',
     sa.Column('user_id', sa.String(length=64), nullable=False),
     sa.Column('content', sa.Text(), nullable=False),
@@ -32,12 +31,9 @@ def upgrade(name: str = "") -> None:
     sa.PrimaryKeyConstraint('user_id', name=op.f('pk_amrita_plugin_memory_user_memo')),
     info={'bind_key': 'amrita_plugin_memory'}
     )
-    # ### end Alembic commands ###
 
 
 def downgrade(name: str = "") -> None:
     if name:
         return
-    # ### commands auto generated by Alembic - please adjust! ###
     op.drop_table('amrita_plugin_memory_user_memo')
-    # ### end Alembic commands ###
diff --git a/amrita_plugin_memory/migrations/6004d221a7de_state.py b/amrita_plugin_memory/migrations/6004d221a7de_state.py
index 925a5c8..4040eba 100644
--- a/amrita_plugin_memory/migrations/6004d221a7de_state.py
+++ b/amrita_plugin_memory/migrations/6004d221a7de_state.py
@@ -22,7 +22,6 @@
 def upgrade(name: str = "") -> None:
     if name:
         return
-    # ### commands auto generated by Alembic - please adjust! ###
     op.create_table('amrita_plugin_memory_subconscious_state',
     sa.Column('uid', sa.String(length=64), nullable=False),
     sa.Column('payload', sa.Text(), nullable=False),
@@ -30,12 +29,9 @@ def upgrade(name: str = "") -> None:
     sa.PrimaryKeyConstraint('uid', name=op.f('pk_amrita_plugin_memory_subconscious_state')),
     info={'bind_key': 'amrita_plugin_memory'}
     )
-    # ### end Alembic commands ###
 
 
 def downgrade(name: str = "") -> None:
     if name:
         return
-    # ### commands auto generated by Alembic - please adjust! ###
     op.drop_table('amrita_plugin_memory_subconscious_state')
-    # ### end Alembic commands ###
diff --git a/amrita_plugin_memory/rethinking/knowledge.py b/amrita_plugin_memory/rethinking/knowledge.py
index 2cd16a3..7bb6a52 100644
--- a/amrita_plugin_memory/rethinking/knowledge.py
+++ b/amrita_plugin_memory/rethinking/knowledge.py
@@ -54,8 +54,6 @@ def __init__(
         self._collection: Collection | None = None
         self._index: list[KnowledgeEntry] = []
 
-    #   生命周期
-
     async def init(self) -> None:
         """创建目录，获取 ChromaDB collection。"""
         self._knowledge_dir.mkdir(parents=True, exist_ok=True)
@@ -76,7 +74,6 @@ async def validate_on_startup(self) -> None:
         index = self._load_index()
         index_map: dict[str, KnowledgeEntry] = {e["kid"]: e for e in index}
 
-        # 扫描 knowledge/ 目录
         existing_files: set[str] = set()
         if self._knowledge_dir.exists():
             for f in self._knowledge_dir.iterdir():
@@ -125,8 +122,6 @@ async def validate_on_startup(self) -> None:
                 f"[KB] validate_on_startup: all {len(index)} entries consistent"
             )
 
-    #   公开 API
-
     async def list_all(self) -> list[KnowledgeListItem]:
         """返回索引中全部知识条目（不含正文）。"""
         return [
@@ -304,7 +299,6 @@ async def search(self, query: str, top_k: int = 5) -> list[KnowledgeSearchItem]:
             list[dict[str, object]], (raw_result.get("metadatas") or [[]])[0]
         )
 
-        # 从索引获取完整信息
         index_map = {e["kid"]: e for e in self._index}
         items: list[KnowledgeSearchItem] = []
         for i, kid in enumerate(ids):
@@ -322,8 +316,6 @@ async def search(self, query: str, top_k: int = 5) -> list[KnowledgeSearchItem]:
             )
         return items
 
-    #   内部方法
-
     async def _recover_orphan_file(
         self,
         kid: str,
diff --git a/amrita_plugin_memory/rethinking/runner.py b/amrita_plugin_memory/rethinking/runner.py
index 45477ba..7061bec 100644
--- a/amrita_plugin_memory/rethinking/runner.py
+++ b/amrita_plugin_memory/rethinking/runner.py
@@ -81,7 +81,6 @@ def __init__(self, config: SubconsciousConfig) -> None:
         self._kb_manager: KnowledgeBaseManager | None = None
         # session 摘要缓存（LRU）：session DB id -> 摘要文本，最多 128 条
         self._session_cache: LRUCache[int, str] = LRUCache(128)
-        # 用户画像文件
         self._profile_path = DATA_PATH / "user_profile.md"
 
     @property
@@ -95,8 +94,6 @@ def _build_config(self) -> AmritaConfig:
         cfg.llm.enable_memory_abstract = self._config.enable_memory_compress
         return cfg
 
-    #  生命周期
-
     async def start(self) -> None:
         logger.info(f"[Subconscious] Starting for user={self._config.target_user_id}")
         await self._load_state()
@@ -150,8 +147,6 @@ def _schedule_once(self, delay_seconds: int) -> None:
             misfire_grace_time=30,
         )
 
-    #  核心运行
-
     async def _run(self) -> None:
         """调度入口 — 保证任何异常路径都会释放运行标志。"""
         if self._is_running:
@@ -210,8 +205,6 @@ def _is_native_thinking(preset: ModelPreset) -> bool:
             and preset.thinking_config.thinking_type == "enabled"
         )
 
-    #  后处理
-
     async def _post_process(self, chat_obj: ChatObject) -> None:
         """本轮结束后：提取摘要、更新全局 usage、持久化、调度下次运行。"""
         # 1. 更新全局 usage（复用 Bot 的 InsightsModel 统计）
@@ -275,8 +268,6 @@ async def _update_global_usage(chat_obj: ChatObject) -> None:
         except Exception as e:
             logger.warning(f"[Subconscious] Update global usage failed: {e}")
 
-    #  SubconsciousState 持久化
-
     @staticmethod
     async def _read_state_payload(uid: str) -> dict[str, Any]:
         async with get_session() as session:
@@ -376,8 +367,6 @@ async def _save_state(self) -> None:
             },
         )
 
-    #  Prompt 加载
-
     async def _load_prompt(self) -> str:
         main_path = (self._prompt_dir / self._config.prompt_file).resolve()
         kn_path = (self._prompt_dir / self._config.prompt_knowledge_file).resolve()
@@ -421,7 +410,7 @@ async def _load_prompt(self) -> str:
                 f"审查后对值得保留的用 subconscious_knowledge_create/update 实际写入。"
             )
 
-        # Phase 3: 膨胀感知 — 查 ChromaDB 总量，超阈值注入警告
+        # 膨胀感知：查 ChromaDB 总量，超阈值注入警告
         try:
             pid = make_scope_id(self._config.target_user_id, is_group=False)
             ope = AsyncUserMemory(get_db_conn())
@@ -447,8 +436,6 @@ async def _load_prompt(self) -> str:
 
         return prompt
 
-    #  Session 读取 & 摘要缓存
-
     async def _read_recent_sessions(self, n: int = 5) -> list[SessionSummary]:
         """读取目标用户最近 N 个归档 sessions，按需生成摘要并缓存。"""
         try:
@@ -530,8 +517,6 @@ async def _summarize_session(self, session: MemorySessionsSchema) -> str:
                 else f"[{date_str}] 无法生成摘要"
             )
 
-    #  用户画像
-
     async def _read_profile(
         self, start_line: int | None = None, end_line: int | None = None
     ) -> ProfileResult:
@@ -598,11 +583,9 @@ async def _update_profile(
         body_lines = existing_body.split("\n") if existing_body else []
         new = new_lines.split("\n")
         if start_line is None or end_line is None:
-            # 追加模式
             new_body_lines = body_lines + new
             operation = f"append {len(new)} lines"
         else:
-            # 替换模式
             s = max(0, start_line)
             e = max(s, min(len(body_lines), end_line))
             new_body_lines = body_lines[:s] + new + body_lines[e:]
diff --git a/amrita_plugin_memory/rethinking/schemas.py b/amrita_plugin_memory/rethinking/schemas.py
index 0c61ef1..9c8f723 100644
--- a/amrita_plugin_memory/rethinking/schemas.py
+++ b/amrita_plugin_memory/rethinking/schemas.py
@@ -145,7 +145,6 @@
     ),
 )
 
-# 工具注册常量
 
 DUPLICATE_HELPER_SCHEMA = FunctionDefinitionSchema(
     name="subconscious_duplicate_helper",
@@ -183,7 +182,6 @@
     parameters=FunctionParametersSchema(type="object", properties={}, required=[]),
 )
 
-#  全局知识库工具
 
 KNOWLEDGE_LIST_SCHEMA = FunctionDefinitionSchema(
     name="subconscious_knowledge_list",
@@ -281,7 +279,7 @@
     ),
 )
 
-#  知识建议（对话 LLM 提议，潜意识 Agent 审查后实际写入）
+# 知识建议：对话 LLM 提议，潜意识 Agent 审查后实际写入
 
 KNOWLEDGE_SUGGEST_SCHEMA = FunctionDefinitionSchema(
     name="knowledge_suggest",
@@ -328,7 +326,6 @@
     parameters=FunctionParametersSchema(type="object", properties={}, required=[]),
 )
 
-#  Session 与用户画像工具
 
 READ_SESSIONS_SCHEMA = FunctionDefinitionSchema(
     name="subconscious_read_sessions",
diff --git a/amrita_plugin_memory/rethinking/tools.py b/amrita_plugin_memory/rethinking/tools.py
index 16ba86b..b243237 100644
--- a/amrita_plugin_memory/rethinking/tools.py
+++ b/amrita_plugin_memory/rethinking/tools.py
@@ -43,8 +43,6 @@
 
 _SUBCONSCIOUS_TOOLS = _state.get_tools_manager()
 
-#  辅助
-
 
 def _make_operator() -> AsyncUserMemory:
     return AsyncUserMemory(get_db_conn())
@@ -67,9 +65,6 @@ def _tools_ok(**extra: Any) -> str:
     return json.dumps({"status": "success", **extra}, ensure_ascii=False, indent=2)
 
 
-#  Handler
-
-
 @on_tools(READ_MEMORY_SCHEMA, strict=True, bound_to=_SUBCONSCIOUS_TOOLS)
 async def subconscious_read_memory(data: dict[str, Any]) -> str:
     ope = _make_operator()
@@ -259,7 +254,6 @@ async def subconscious_send_to_user(data: dict[str, Any]) -> str:
     pending = _state.get_pending()
     pending.append({"content": content, "timestamp": ts})
     await runner._save_state()
-    # 即时发送
     try:
         from nonebot import get_bot
 
@@ -306,9 +300,6 @@ async def subconscious_read_chat_context(data: dict[str, Any]) -> str:
         return _tools_err(str(e))
 
 
-#  消息生成
-
-
 async def _generate_send_content(intent: str, memory_context: str) -> str:
     runner = _state.get_runner()
     if runner is None:
@@ -337,9 +328,6 @@ async def _generate_send_content(intent: str, memory_context: str) -> str:
     return response.content.strip()
 
 
-#  压缩辅助工具
-
-
 @on_tools(DUPLICATE_HELPER_SCHEMA, strict=True, bound_to=_SUBCONSCIOUS_TOOLS)
 async def subconscious_duplicate_helper(data: dict[str, Any]) -> str:
     """返回指定范围内全部记忆 + LLM 合并指导 prompt。
@@ -364,7 +352,6 @@ async def subconscious_duplicate_helper(data: dict[str, Any]) -> str:
             {} for _ in ids
         ]
 
-        # 过滤
         tag_filter = data.get("tag")
         imp_filter = data.get("importance")
         sort_by = str(data.get("sort_by", "created_at"))
@@ -387,7 +374,6 @@ async def subconscious_duplicate_helper(data: dict[str, Any]) -> str:
                 }
             )
 
-        # 排序
         if sort_by == "importance":
             importance_order = {"high": 0, "medium": 1, "low": 2}
             items.sort(key=lambda x: importance_order.get(str(x["importance"]), 1))
@@ -482,9 +468,6 @@ async def subconscious_get_memory_stats(data: dict[str, Any]) -> str:
         return _tools_err(str(e))
 
 
-#  全局知识库工具
-
-
 def _get_kb_manager():
     """获取 KnowledgeBaseManager 实例，未初始化或已禁用则报错。"""
     runner = _state.get_runner()
@@ -622,7 +605,7 @@ async def subconscious_knowledge_search(data: dict[str, Any]) -> str:
         return _tools_err(str(e))
 
 
-#  知识建议（对话 LLM -> 潜意识 Agent 审查管线）
+# 知识建议：对话 LLM -> 潜意识 Agent 审查管线
 
 
 @on_tools(KNOWLEDGE_SUGGEST_SCHEMA, strict=True)
@@ -676,9 +659,6 @@ async def subconscious_read_suggestions(data: dict[str, Any]) -> str:
     )
 
 
-#  Session 与用户画像工具
-
-
 def _get_runner():
     """获取 SubconsciousRunner 实例，未初始化则报错。"""
     runner = _state.get_runner()
diff --git a/amrita_plugin_memory/tools.py b/amrita_plugin_memory/tools.py
index 4442f5b..39929ce 100644
--- a/amrita_plugin_memory/tools.py
+++ b/amrita_plugin_memory/tools.py
@@ -62,7 +62,6 @@ def _check_required(ctx: ToolContext, tool_name: str, *params: str) -> str | Non
     return None
 
 
-#  公共参数：scope
 _SCOPE_PROP = FunctionPropertySchema(
     type="string",
     description=(
@@ -72,7 +71,6 @@ def _check_required(ctx: ToolContext, tool_name: str, *params: str) -> str | Non
     enum=["group", "user"],
 )
 
-#  Function Schema 定义
 
 WRITE_MEMORY_FUN = FunctionDefinitionSchema(
     name="write_memory",
@@ -199,9 +197,6 @@ def _check_required(ctx: ToolContext, tool_name: str, *params: str) -> str | Non
 )
 
 
-#  Handler 实现
-
-
 @on_tools(WRITE_MEMORY_FUN, custom_run=True, strict=True)
 async def w(ctx: ToolContext) -> str:
     if err := _check_required(