diff --git a/server/deep_compression.py b/server/deep_compression.py
index 0fc346d..23f5888 100644
--- a/server/deep_compression.py
+++ b/server/deep_compression.py
@@ -19,7 +19,7 @@ SUMMARY_PROMPT = (
     "请使用中文，结构清晰，尽量具体，不要省略关键上下文。"
 )
 
-GUIDE_USER_MESSAGE_TEMPLATE = "当前对话已经被自动压缩，请阅读{path}并继续工作"
+GUIDE_USER_MESSAGE_TEMPLATE = "当前对话已经被自动压缩（第{count}次）。请阅读{path}并继续工作"
 
 
 def _emit(sender, event_type: str, payload: Dict[str, Any]):
@@ -31,6 +31,54 @@ def _emit(sender, event_type: str, payload: Dict[str, Any]):
         pass
 
 
+def _normalize_deep_compression_records(metadata: Dict[str, Any]) -> List[Dict[str, Any]]:
+    records = metadata.get("deep_compression_records")
+    if not isinstance(records, list):
+        return []
+    normalized: List[Dict[str, Any]] = []
+    for item in records:
+        if not isinstance(item, dict):
+            continue
+        try:
+            count = int(item.get("count", 0) or 0)
+        except Exception:
+            count = 0
+        path = str(item.get("compact_file") or "").strip()
+        if count <= 0 or not path:
+            continue
+        normalized.append({
+            "count": count,
+            "compact_file": path,
+            "created_at": item.get("created_at"),
+            "source_conversation_id": item.get("source_conversation_id"),
+            "compressed_conversation_id": item.get("compressed_conversation_id"),
+        })
+    normalized.sort(key=lambda x: (int(x.get("count") or 0), str(x.get("created_at") or "")))
+    deduped: List[Dict[str, Any]] = []
+    seen = set()
+    for rec in normalized:
+        key = (int(rec.get("count") or 0), str(rec.get("compact_file") or ""))
+        if key in seen:
+            continue
+        seen.add(key)
+        deduped.append(rec)
+    return deduped
+
+
+def _build_guide_message(*, compression_index: int, compact_file: str, previous_records: List[Dict[str, Any]]) -> str:
+    base = GUIDE_USER_MESSAGE_TEMPLATE.format(count=compression_index, path=compact_file)
+    if not previous_records:
+        return base
+    lines = [base, "", "此前压缩摘要文件位置："]
+    for rec in previous_records:
+        count = int(rec.get("count") or 0)
+        path = str(rec.get("compact_file") or "").strip()
+        if count <= 0 or not path:
+            continue
+        lines.append(f"- 第{count}次：{path}")
+    return "\n".join(lines).strip()
+
+
 def _extract_text_only(content: Any) -> str:
     if content is None:
         return ""
@@ -202,7 +250,9 @@ async def run_deep_compression(
 
     old_title = conv_data.get("title") or "未命名"
     compression_count = int(metadata.get("compression_count", 0) or 0)
-    target_count = compression_count + 1
+    previous_records = _normalize_deep_compression_records(metadata)
+    previous_max_count = max([int(item.get("count") or 0) for item in previous_records], default=0)
+    target_count = max(compression_count, previous_max_count) + 1
     job_id = f"cmp_{datetime.now().strftime('%Y%m%d_%H%M%S_%f')}"
     cm.set_compression_state(
         in_progress=True,
@@ -271,9 +321,29 @@ async def run_deep_compression(
             "skip_auto_title_generation": True,
         }
     )
+    current_record = {
+        "count": target_count,
+        "compact_file": relative_compact_path,
+        "created_at": datetime.now().isoformat(),
+        "source_conversation_id": conversation_id,
+        "compressed_conversation_id": new_conv_id,
+    }
+    all_records = previous_records + [current_record]
+    cm.conversation_manager.update_conversation_metadata(
+        new_conv_id,
+        {
+            "compression_count": target_count,
+            "deep_compression_records": all_records,
+            "last_deep_compression_record": current_record,
+        },
+    )
     cm.conversation_manager.update_conversation_title(new_conv_id, f"{old_title}对话 压缩后")
     web_terminal.load_conversation(new_conv_id)
-    guide_message = GUIDE_USER_MESSAGE_TEMPLATE.format(path=relative_compact_path)
+    guide_message = _build_guide_message(
+        compression_index=target_count,
+        compact_file=relative_compact_path,
+        previous_records=previous_records,
+    )
 
     # 清理旧对话压缩状态
     cm.conversation_manager.update_conversation_metadata(
diff --git a/utils/conversation_manager.py b/utils/conversation_manager.py
index b11143e..a3ad417 100644
--- a/utils/conversation_manager.py
+++ b/utils/conversation_manager.py
@@ -395,6 +395,8 @@ class ConversationManager:
             "compression_job_id": None,
             "compression_error": None,
             "compression_resume_payload": None,
+            "deep_compression_records": [],
+            "last_deep_compression_record": None,
             "skip_auto_title_generation": False,
         }
         if isinstance(metadata_overrides, dict):