Dev v2.0.14

.gitignore

@@ -67,6 +67,8 @@ report/
 cov-report/
 .tox/
 .nox/
+report/
+cov-report/
 .coverage
 .coverage.*
 .cache

src/memos/api/config.py

@@ -369,6 +369,26 @@ def get_memreader_config() -> dict[str, Any]:
 
         return {"backend": "openai", "config": config}
 
+    @staticmethod
+    def get_qwen_llm_config() -> dict[str, Any] | None:
+        if not os.getenv("QWEN_API_KEY"):
+            return None
+        return {
+            "backend": "qwen",
+            "config": {
+                "model_name_or_path": os.getenv("QWEN_MODEL", "qwen-flash"),
+                "temperature": float(os.getenv("QWEN_TEMPERATURE", "0.8")),
+                "max_tokens": int(os.getenv("QWEN_MAX_TOKENS", "8000")),
+                "top_p": float(os.getenv("QWEN_TOP_P", "0.9")),
+                "top_k": int(os.getenv("QWEN_TOP_K", "50")),
+                "remove_think_prefix": os.getenv("QWEN_REMOVE_THINK_PREFIX", "true").lower()
+                == "true",
+                "api_key": os.getenv("QWEN_API_KEY", ""),
+                "api_base": os.getenv("QWEN_API_BASE", ""),
+                "model_schema": os.getenv("QWEN_MODEL_SCHEMA", "memos.configs.llm.QwenLLMConfig"),
+            },
+        }
+
     @staticmethod
     def get_memreader_general_llm_config() -> dict[str, Any]:
         """Get general LLM configuration for non-chat/doc tasks.
@@ -639,6 +659,7 @@ def get_oss_config() -> dict[str, Any] | None:
 
         return config
 
+    @staticmethod
     def get_internet_config() -> dict[str, Any]:
         """Get internet retriever configuration.
 
@@ -705,8 +726,9 @@ def get_internet_config() -> dict[str, Any]:
 
     @staticmethod
     def get_nli_config() -> dict[str, Any]:
-        """Get NLI model configuration."""
+        """Get relation-judge configuration for memory-version candidate matching."""
         return {
+            "provider": os.getenv("MEM_VERSION_RELATION_JUDGE_PROVIDER", "llm"),
             "base_url": os.getenv("NLI_MODEL_BASE_URL", "http://localhost:32532"),
         }
 
@@ -872,7 +894,7 @@ def get_scheduler_config() -> dict[str, Any]:
                 ),
                 "context_window_size": int(os.getenv("MOS_SCHEDULER_CONTEXT_WINDOW_SIZE", "5")),
                 "thread_pool_max_workers": int(
-                    os.getenv("MOS_SCHEDULER_THREAD_POOL_MAX_WORKERS", "200")
+                    os.getenv("MOS_SCHEDULER_THREAD_POOL_MAX_WORKERS", "50")
                 ),
                 "consume_interval_seconds": float(
                     os.getenv("MOS_SCHEDULER_CONSUME_INTERVAL_SECONDS", "0.01")
@@ -952,6 +974,7 @@ def get_product_default_config() -> dict[str, Any]:
                 "backend": reader_config["backend"],
                 "config": {
                     "llm": APIConfig.get_memreader_config(),
+                    "qwen_llm": APIConfig.get_qwen_llm_config(),
                     # General LLM for non-chat/doc tasks (hallucination filter, rewrite, merge, etc.)
                     "general_llm": APIConfig.get_memreader_general_llm_config(),
                     # Image parser LLM (requires vision model)
@@ -986,6 +1009,7 @@ def get_product_default_config() -> dict[str, Any]:
                             "SKILLS_LOCAL_DIR", "/tmp/upload_skill_memory/"
                         ),
                     },
+                    "memory_version_switch": os.getenv("MEM_READER_MEM_VERSION_SWITCH", "off"),
                 },
             },
             "enable_textual_memory": True,

src/memos/api/handlers/add_handler.py

@@ -15,6 +15,7 @@
 from memos.multi_mem_cube.composite_cube import CompositeCubeView
 from memos.multi_mem_cube.single_cube import SingleCubeView
 from memos.multi_mem_cube.views import MemCubeView
+from memos.plugins.hooks import hookable
 from memos.types import MessageList
 
 
@@ -37,6 +38,7 @@ def __init__(self, dependencies: HandlerDependencies):
             "naive_mem_cube", "mem_reader", "mem_scheduler", "feedback_server"
         )
 
+    @hookable("add")
     def handle_add_memories(self, add_req: APIADDRequest) -> MemoryResponse:
         """
         Main handler for add memories endpoint.

src/memos/api/handlers/component_init.py

@@ -32,14 +32,14 @@
 from memos.mem_scheduler.orm_modules.base_model import BaseDBManager
 from memos.mem_scheduler.scheduler_factory import SchedulerFactory
 from memos.memories.textual.simple_tree import SimpleTreeTextMemory
-from memos.memories.textual.tree_text_memory.organize.history_manager import MemoryHistoryManager
 from memos.memories.textual.tree_text_memory.organize.manager import MemoryManager
 from memos.memories.textual.tree_text_memory.retrieve.retrieve_utils import FastTokenizer
+from memos.plugins.component_bootstrap import build_plugin_context
+from memos.plugins.manager import plugin_manager
 
 
 if TYPE_CHECKING:
     from memos.memories.textual.tree import TreeTextMemory
-from memos.extras.nli_model.client import NLIClient
 from memos.mem_agent.deepsearch_agent import DeepSearchMemAgent
 from memos.memories.textual.tree_text_memory.retrieve.internet_retriever_factory import (
     InternetRetrieverFactory,
@@ -122,6 +122,12 @@ def init_server() -> dict[str, Any]:
         existing code that uses the components.
     """
     logger.info("Initializing MemOS server components...")
+    logger.info(
+        "[INIT_SERVER] env_MEMSCHEDULER_STREAM_KEY_PREFIX=%s, env_MEMSCHEDULER_REDIS_STREAM_KEY_PREFIX=%s, env_POLAR_DB_DB_NAME=%s",
+        os.getenv("MEMSCHEDULER_STREAM_KEY_PREFIX"),
+        os.getenv("MEMSCHEDULER_REDIS_STREAM_KEY_PREFIX"),
+        os.getenv("POLAR_DB_DB_NAME"),
+    )
 
     # Initialize Redis client first as it is a core dependency for features like scheduler status tracking
     if os.getenv("MEMSCHEDULER_USE_REDIS_QUEUE", "False").lower() == "true":
@@ -169,10 +175,25 @@ def init_server() -> dict[str, Any]:
         else None
     )
     embedder = EmbedderFactory.from_config(embedder_config)
-    nli_client = NLIClient(base_url=nli_client_config["base_url"])
-    memory_history_manager = MemoryHistoryManager(nli_client=nli_client, graph_db=graph_db)
+
+    plugin_context = build_plugin_context(
+        graph_db=graph_db,
+        embedder=embedder,
+        default_cube_config=default_cube_config,
+        nli_client_config=nli_client_config,
+        mem_reader_config=mem_reader_config,
+        reranker_config=reranker_config,
+        feedback_reranker_config=feedback_reranker_config,
+        internet_retriever_config=internet_retriever_config,
+    )
+    plugin_manager.discover()
+    plugin_manager.init_components(plugin_context)
+
     # Pass graph_db to mem_reader for recall operations (deduplication, conflict detection)
-    mem_reader = MemReaderFactory.from_config(mem_reader_config, graph_db=graph_db)
+    mem_reader = MemReaderFactory.from_config(
+        mem_reader_config,
+        graph_db=graph_db,
+    )
     reranker = RerankerFactory.from_config(reranker_config)
     feedback_reranker = RerankerFactory.from_config(feedback_reranker_config)
     internet_retriever = InternetRetrieverFactory.from_config(
@@ -303,6 +324,4 @@ def init_server() -> dict[str, Any]:
         "feedback_server": feedback_server,
         "redis_client": redis_client,
         "deepsearch_agent": deepsearch_agent,
-        "nli_client": nli_client,
-        "memory_history_manager": memory_history_manager,
     }

src/memos/api/handlers/formatters_handler.py

@@ -141,6 +141,7 @@ def separate_knowledge_and_conversation_mem(memories: list[dict[str, Any]]):
         sources = item.get("metadata", {}).get("sources", [])
         if (
             item["metadata"]["memory_type"] != "RawFileMemory"
+            and sources
             and len(sources) > 0
             and "type" in sources[0]
             and sources[0]["type"] == "file"

src/memos/api/server_api.py

@@ -9,13 +9,21 @@
 from memos.api.exceptions import APIExceptionHandler
 from memos.api.middleware.request_context import RequestContextMiddleware
 from memos.api.routers.server_router import router as server_router
+from memos.plugins.manager import plugin_manager
 
 
 load_dotenv()
 
+plugin_manager.discover()
+
 # Configure logging
 logging.basicConfig(level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s")
 logger = logging.getLogger(__name__)
+logger.info(
+    "[SERVER_API] load_dotenv completed. env_MEMSCHEDULER_STREAM_KEY_PREFIX=%s, env_MEMSCHEDULER_REDIS_STREAM_KEY_PREFIX=%s",
+    os.getenv("MEMSCHEDULER_STREAM_KEY_PREFIX"),
+    os.getenv("MEMSCHEDULER_REDIS_STREAM_KEY_PREFIX"),
+)
 
 app = FastAPI(
     title="MemOS Server REST APIs",
@@ -49,6 +57,8 @@ def health_check():
 # Fallback for unknown errors
 app.exception_handler(Exception)(APIExceptionHandler.global_exception_handler)
 
+plugin_manager.init_app(app)
+
 
 if __name__ == "__main__":
     import argparse

src/memos/chunkers/sentence_chunker.py

@@ -53,4 +53,5 @@ def chunk(self, text: str) -> list[str] | list[Chunk]:
             chunks.append(chunk)
 
         logger.debug(f"Generated {len(chunks)} chunks from input text")
+
         return chunks

src/memos/configs/mem_reader.py

@@ -1,5 +1,5 @@
 from datetime import datetime
-from typing import Any, ClassVar
+from typing import Any, ClassVar, Literal
 
 from pydantic import ConfigDict, Field, field_validator, model_validator
 
@@ -76,6 +76,13 @@ class MultiModalStructMemReaderConfig(BaseMemReaderConfig):
         default=None,
         description="Skills directory for the MemReader",
     )
+    memory_version_switch: Literal["on", "off"] = Field(
+        default="off",
+        description="Turn on memory version or off",
+    )
+
+    # Allow passing additional fields without raising validation errors
+    model_config = ConfigDict(extra="allow", strict=True)
 
 
 class StrategyStructMemReaderConfig(BaseMemReaderConfig):