chore: bump version to v2.0.0

.github/workflows/python-tests.yml

@@ -28,7 +28,6 @@ jobs:
         os:
           - "ubuntu-latest"
           - "windows-latest"
-          - "macos-13"
           - "macos-14"
           - "macos-15"
           # Ref: https://docs.github.com/en/actions/how-tos/writing-workflows/choosing-where-your-workflow-runs/choosing-the-runner-for-a-job

README.md

@@ -3,13 +3,15 @@
 MemOS is an open-source **Agent Memory framework** that empowers AI agents with **long-term memory, personality consistency, and contextual recall**. It enables agents to **remember past interactions**, **learn over time**, and **build evolving identities** across sessions.
 
 Designed for **AI companions, role-playing NPCs, and multi-agent systems**, MemOS provides a unified API for **memory representation, retrieval, and update** — making it the foundation for next-generation **memory-augmented AI agents**.
+
+🆕 **MemOS 2.0** introduces **knowledge base system**, **multi-modal memory** (images & documents), **tool memory** for Agent optimization, **memory feedback mechanism** for precise control, and **enterprise-grade architecture** with Redis Streams scheduler and advanced DB optimizations.
 <div align="center">
   <a href="https://memos.openmem.net/">
     <img src="https://statics.memtensor.com.cn/memos/memos-banner.gif" alt="MemOS Banner">
   </a>
 
 <h1 align="center">
-  <img src="https://statics.memtensor.com.cn/logo/memos_color_m.png" alt="MemOS Logo" width="50"/> MemOS 1.0: 星河 (Stellar)  <img src="https://img.shields.io/badge/status-Preview-blue" alt="Preview Badge"/>
+  <img src="https://statics.memtensor.com.cn/logo/memos_color_m.png" alt="MemOS Logo" width="50"/> MemOS 2.0: 星尘（Stardust） <img src="https://img.shields.io/badge/status-Preview-blue" alt="Preview Badge"/>
 </h1>
 
   <p>
@@ -60,7 +62,7 @@ Get Free API: [Try API](https://memos-dashboard.openmem.net/quickstart/?source=g
 
 <img src="https://cdn.memtensor.com.cn/img/1762436050812_3tgird_compressed.png" alt="SOTA SCORE">
 
-**MemOS** is an operating system for Large Language Models (LLMs) that enhances them with long-term memory capabilities. It allows LLMs to store, retrieve, and manage information, enabling more context-aware, consistent, and personalized interactions.
+**MemOS** is an operating system for Large Language Models (LLMs) that enhances them with long-term memory capabilities. It allows LLMs to store, retrieve, and manage information, enabling more context-aware, consistent, and personalized interactions. **MemOS 2.0** features comprehensive knowledge base management, multi-modal memory support, tool memory for Agent enhancement, and enterprise-grade architecture optimizations.
 
 - **Website**: https://memos.openmem.net/
 - **Documentation**: https://memos-docs.openmem.net/home/overview/
@@ -71,7 +73,8 @@ Get Free API: [Try API](https://memos-dashboard.openmem.net/quickstart/?source=g
 
 Stay up to date with the latest MemOS announcements, releases, and community highlights.
 
-
+- **2025-12-24** - 🎉 **MemOS v2.0: Stardust (星尘) Release**:
+  Major upgrade featuring comprehensive Knowledge Base system with automatic document/URL parsing and cross-project sharing; Memory feedback mechanism for correction and precise deletion; Multi-modal memory supporting images and charts; Tool Memory to enhance Agent planning; Full architecture upgrade with Redis Streams multi-level queue scheduler and DB optimizations; New streaming/non-streaming Chat interfaces; Complete MCP upgrade; Lightweight deployment modes (quick & full).
 - **2025-11-06** - 🎉 MemOS v1.1.3 (Async Memory & Preference):
   Millisecond-level async memory add (support plain-text-memory and
   preference memory); enhanced BM25, graph recall, and mixture search; full
@@ -114,7 +117,19 @@ showcasing its capabilities in **information extraction**, **temporal and cross-
     - **Textual Memory**: For storing and retrieving unstructured or structured text knowledge.
     - **Activation Memory**: Caches key-value pairs (`KVCacheMemory`) to accelerate LLM inference and context reuse.
     - **Parametric Memory**: Stores model adaptation parameters (e.g., LoRA weights).
+    - **Tool Memory** 🆕: Records Agent tool call trajectories and experiences to improve planning capabilities.
+- **📚 Knowledge Base System** 🆕: Build multi-dimensional knowledge bases with automatic document/URL parsing, splitting, and cross-project sharing capabilities.
+- **🔧 Memory Controllability** 🆕:
+    - **Feedback Mechanism**: Use `add_feedback` API to correct, supplement, or replace existing memories with natural language.
+    - **Precise Deletion**: Delete specific memories by User ID or Memory ID via API or MCP tools.
+- **👁️ Multi-Modal Support** 🆕: Support for image understanding and memory, including chart parsing in documents.
+- **⚡ Advanced Architecture**:
+    - **DB Optimization**: Enhanced connection management and batch insertion for high-concurrency scenarios.
+    - **Advanced Retrieval**: Custom tag and info field filtering with complex logical operations.
+    - **Redis Streams Scheduler**: Multi-level queue architecture with intelligent orchestration for fair multi-tenant scheduling.
+    - **Stream & Non-Stream Chat**: Ready-to-use streaming and non-streaming chat interfaces.
 - **🔌 Extensible**: Easily extend and customize memory modules, data sources, and LLM integrations.
+- **🏂 Lightweight Deployment** 🆕: Support for quick mode and complete mode deployment options.
 
 ## 🚀 Getting Started
 

docker/.env.example

@@ -47,7 +47,7 @@ OLLAMA_API_BASE=http://localhost:11434     # required when backend=ollama
 MOS_RERANKER_BACKEND=http_bge             # http_bge | http_bge_strategy | cosine_local
 MOS_RERANKER_URL=http://localhost:8001     # required when backend=http_bge*
 MOS_RERANKER_MODEL=bge-reranker-v2-m3     # siliconflow → use BAAI/bge-reranker-v2-m3
-MOS_RERANKER_HEADERS_EXTRA=               # extra headers, JSON string
+MOS_RERANKER_HEADERS_EXTRA=               # extra headers, JSON string, e.g. {"Authorization":"Bearer your_token"}
 MOS_RERANKER_STRATEGY=single_turn
 MOS_RERANK_SOURCE=                        # optional rerank scope, e.g., history/stream/custom
 
@@ -93,6 +93,9 @@ NEO4J_DB_NAME=neo4j                       # required for shared-db mode
 MOS_NEO4J_SHARED_DB=false
 QDRANT_HOST=localhost
 QDRANT_PORT=6333
+# For Qdrant Cloud / remote endpoint (takes priority if set):
+QDRANT_URL=your_qdrant_url
+QDRANT_API_KEY=your_qdrant_key
 MILVUS_URI=http://localhost:19530         # required when ENABLE_PREFERENCE_MEMORY=true
 MILVUS_USER_NAME=root                     # same as above
 MILVUS_PASSWORD=12345678                  # same as above
@@ -164,11 +167,6 @@ OSS_ACCESS_KEY_ID=
 OSS_ACCESS_KEY_SECRET=
 OSS_PUBLIC_BASE_URL=
 
-## Logging / external sink
-CUSTOM_LOGGER_URL=
-CUSTOM_LOGGER_TOKEN=
-CUSTOM_LOGGER_WORKERS=2
-
 ## SDK / external client
 MEMOS_API_KEY=
 MEMOS_BASE_URL=https://memos.memtensor.cn/api/openmem/v1

docker/requirements.txt

@@ -159,3 +159,7 @@ watchfiles==1.1.0
 websockets==15.0.1
 xlrd==2.0.2
 xlsxwriter==3.2.5
+prometheus-client==0.23.1
+pymilvus==2.5.12
+nltk==3.9.1
+rake-nltk==1.0.6

docs/openapi.json

@@ -884,7 +884,7 @@
             "type": "string",
             "title": "Session Id",
             "description": "Session ID for the MOS. This is used to distinguish between different dialogue",
-            "default": "41bb5e18-252d-4948-918c-07d82aa47086"
+            "default": "8dcdbd62-c231-4678-a3ae-0946b7d9ce14"
           },
           "chat_model": {
             "$ref": "#/components/schemas/LLMConfigFactory",

docs/product-api-tests.md

@@ -0,0 +1,65 @@
+## Product API smoke tests (local 0.0.0.0:8001)
+
+Source: https://github.com/MemTensor/MemOS/issues/518
+
+### Prerequisites
+- Service is running: `python -m uvicorn memos.api.server_api:app --host 0.0.0.0 --port 8001`
+- `.env` is configured for Redis, embeddings, and the vector DB (current test setup: Redis reachable, Qdrant Cloud connected).
+
+### 1) /product/add
+- Purpose: Write a memory (sync/async).
+- Example request (sync):
+
+  ```bash
+  curl -s -X POST http://127.0.0.1:8001/product/add \
+    -H 'Content-Type: application/json' \
+    -d '{
+          "user_id": "tester",
+          "mem_cube_id": "default_cube",
+          "memory_content": "Apple is a fruit rich in fiber.",
+          "async_mode": "sync"
+        }'
+  ```
+
+- Observed result: `200`, message: "Memory added successfully", returns the written `memory_id` and related info.
+
+### 2) /product/get_all
+- Purpose: List all memories for the user/type to confirm writes.
+- Example request:
+
+  ```bash
+  curl -s -X POST http://127.0.0.1:8001/product/get_all \
+    -H 'Content-Type: application/json' \
+    -d '{
+          "user_id": "tester",
+          "memory_type": "text_mem",
+          "mem_cube_ids": ["default_cube"]
+        }'
+  ```
+
+- Observed result: `200`, shows the recently written apple memories (WorkingMemory/LongTermMemory/UserMemory present, `vector_sync=success`).
+
+### 3) /product/search
+- Purpose: Vector search memories.
+- Example request:
+
+  ```bash
+  curl -s -X POST http://127.0.0.1:8001/product/search \
+    -H 'Content-Type: application/json' \
+    -d '{
+          "query": "What fruit is rich in fiber?",
+          "user_id": "tester",
+          "mem_cube_id": "default_cube",
+          "top_k": 5,
+          "pref_top_k": 3,
+          "include_preference": false
+        }'
+  ```
+
+- Observed result: previously returned 400 because payload indexes (e.g., `vector_sync`) were missing in Qdrant. Index creation is now automatic during Qdrant initialization (memory_type/status/vector_sync/user_name).
+- If results are empty or errors persist, verify indexes exist (auto-created on restart) or recreate/clean the collection.
+
+### Notes / Next steps
+- `/product/add` and `/product/get_all` are healthy.
+- `/product/search` still returns empty results even with vectors present; likely related to search filters or vector retrieval.
+- Suggested follow-ups: inspect `SearchHandler` flow, filter conditions (user_id/session/cube_name), and vector DB search calls; capture logs or compare with direct `VecDBFactory.search` calls.

evaluation/scripts/locomo/locomo_eval.py

@@ -3,6 +3,7 @@
 import json
 import logging
 import os
+import re
 import time
 
 import nltk
@@ -47,6 +48,29 @@ class LLMGrade(BaseModel):
     llm_reasoning: str = Field(description="Explain why the answer is correct or incorrect.")
 
 
+def extract_label_json(text: str) -> str | None:
+    """
+    Extracts a JSON object of the form {"label": "VALUE"} from a given text string.
+    This function is designed to handle cases where the LLM response contains
+    natural language alongside a final JSON snippet, ensuring robust parsing.
+
+    Supports both single and double quotes around the label value.
+    Ignores surrounding whitespace and formatting.
+
+    Returns:
+        The full matching JSON string (e.g., '{"label": "CORRECT"}') if found.
+        None if no valid label JSON is found.
+    """
+    # Regex pattern to match: { "label": "value" } with optional whitespace
+    # Matches both single and double quotes, allows spaces around keys and values
+    pattern = r'\{\s*"label"\s*:\s*["\']([^"\']*)["\']\s*\}'
+    match = re.search(pattern, text)
+    if match:
+        # Return the complete matched JSON string for safe json.loads()
+        return match.group(0)
+    return None
+
+
 async def locomo_grader(llm_client, question: str, gold_answer: str, response: str) -> bool:
     system_prompt = """
         You are an expert grader that determines if answers to questions match a gold standard answer
@@ -77,20 +101,23 @@ async def locomo_grader(llm_client, question: str, gold_answer: str, response: s
 
     Just return the label CORRECT or WRONG in a json format with the key as "label".
     """
-
-    response = await llm_client.chat.completions.create(
-        model="gpt-4o-mini",
-        messages=[
-            {"role": "system", "content": system_prompt},
-            {"role": "user", "content": accuracy_prompt},
-        ],
-        temperature=0,
-    )
-    message_content = response.choices[0].message.content
-    label = json.loads(message_content)["label"]
-    parsed = LLMGrade(llm_judgment=label, llm_reasoning="")
-
-    return parsed.llm_judgment.strip().lower() == "correct"
+    try:
+        response = await llm_client.chat.completions.create(
+            model=os.getenv("EVAL_MODEL", "gpt-4o-mini"),
+            messages=[
+                {"role": "system", "content": system_prompt},
+                {"role": "user", "content": accuracy_prompt},
+            ],
+            temperature=0,
+        )
+        message_content = response.choices[0].message.content
+        message_content = extract_label_json(text=message_content)
+        label = json.loads(message_content)["label"]
+        parsed = LLMGrade(llm_judgment=label, llm_reasoning="")
+        return parsed.llm_judgment.strip().lower() == "correct"
+    except Exception as e:
+        print(f"======== {e}, {response} ===========")
+        exit()
 
 
 def calculate_rouge_scores(gold_answer, response):

evaluation/scripts/long_bench-v2/__init__.py

@@ -0,0 +1 @@
+# LongBench v2 evaluation scripts