alpha-avatar-plugins-memory 0.6.0__tar.gz → 0.6.2__tar.gz

{alpha_avatar_plugins_memory-0.6.0 → alpha_avatar_plugins_memory-0.6.2}/.gitignore

@@ -179,3 +179,5 @@ pyrightconfig.json
 # Database files
 *.db
 .alphaavatar_workspace
+
+docs/releases/.generated/

alpha_avatar_plugins_memory-0.6.2/PKG-INFO

@@ -0,0 +1,176 @@
+Metadata-Version: 2.4
+Name: alpha-avatar-plugins-memory
+Version: 0.6.2
+Summary: AlphaAvatar Framework plugin for memory service
+Project-URL: Source, https://github.com/AlphaAvatar/AlphaAvatar
+License-Expression: Apache-2.0
+Keywords: AI,agents,langchain,llm,memory
+Classifier: Intended Audience :: Developers
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3 :: Only
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Topic :: Multimedia :: Sound/Audio
+Classifier: Topic :: Multimedia :: Video
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Python: <3.12,>=3.11
+Requires-Dist: lancedb==0.30.2
+Requires-Dist: langchain-community
+Requires-Dist: langchain-core
+Requires-Dist: langchain-qdrant
+Requires-Dist: qdrant-client
+Requires-Dist: torchaudio>=0.10.1
+Description-Content-Type: text/markdown
+
+# 🧠 Memory Plugin for AlphaAvatar
+
+> Give AlphaAvatar the ability to **remember you** — across conversations, sessions, and time.
+
+---
+
+## 🤔 What is the Memory Plugin?
+
+Imagine talking to an assistant that forgets everything you said the moment you close the app. Frustrating, right?
+
+The **Memory Plugin** solves this. It allows AlphaAvatar to:
+- Remember your **name, preferences, and habits**
+- Recall **past conversations** and what was discussed
+- Learn from its own **tool usage** (like web searches or documents it read)
+- Get **smarter and more personal** the more you use it
+
+Think of it like giving AlphaAvatar a notebook — every important thing gets written down and read back when needed.
+
+---
+
+## 💡 How Does it Work? (Simple Flow)
+
+```
+You say something to AlphaAvatar
+            ↓
+AlphaAvatar picks out important information
+(your name, preferences, facts, decisions)
+            ↓
+That information gets saved to local storage
+            ↓
+Next time you start a conversation
+            ↓
+AlphaAvatar reads its saved memories
+            ↓
+Responds like it knows you — because it does
+```
+
+No setup needed from your side. It works automatically in the background.
+
+---
+
+## ✨ Features
+
+### 🌍 Global Memory (Avatar-wide)
+AlphaAvatar maintains memories across your entire experience — not just one chat window.
+
+This includes:
+- Things **you** told it (name, preferences, goals)
+- Things it learned from **external tools** (web searches, documents, research)
+- Things it observed from **its own environment** (what worked, what didn't)
+
+### ⚡ Real-time Context Updates
+Memory doesn't just update at the end of a conversation — it updates **while you're talking**. If you mention something new mid-conversation, it gets remembered immediately.
+
+---
+
+## 📦 What Kind of Things Does AlphaAvatar Remember?
+
+| Type | Real Example |
+|------|-------------|
+| 👤 Personal facts | "My name is Alex and I live in New York" |
+| ❤️ Preferences | "I like short, direct answers" |
+| 🔍 Research results | Facts found during a DeepResearch session |
+| 📄 Document knowledge | Content from files you uploaded via RAG |
+| 🗓️ Past conversations | Topics and decisions from previous sessions |
+| 🛠️ Tool interactions | Results from MCP tools like Notion or Gmail |
+
+---
+
+## 🔧 Installation
+
+```bash
+pip install alpha-avatar-plugins-memory
+```
+
+That's it. The plugin is automatically used when you run AlphaAvatar.
+
+---
+
+## 🗄️ How Memories are Stored
+
+AlphaAvatar uses **vector storage** to save memories. This means memories aren't stored as plain text — they're stored in a way that makes them easy to search and retrieve intelligently.
+
+### Supported Backends
+
+| Backend | What it Does |
+|---------|-------------|
+| **LanceDB** | Saves memories locally on your machine (default) |
+| **Qdrant** | A powerful vector database for storage and retrieval |
+| **LangChain** | The pipeline that extracts and processes memories |
+
+By default, everything is stored **locally** — your memories never leave your machine unless you configure a cloud backend.
+
+---
+
+## 🔗 How Memory Connects to Other Plugins
+
+The Memory plugin doesn't work alone — it's the foundation that other plugins build on:
+
+```
+Memory Plugin
+      ├── → Persona Plugin   (builds your user profile from memories)
+      ├── → RAG Plugin       (stores document knowledge in memory)
+      ├── → Reflection*      (analyzes memories to improve Avatar behavior)
+      └── → Planning*        (uses memories to set reminders and goals)
+
+* Coming soon
+```
+
+---
+
+## 🙋 Common Questions
+
+**Q: Does memory work across different devices?**
+By default, memory is stored locally. Cross-device sync is not yet supported.
+
+**Q: Can I see what AlphaAvatar remembers about me?**
+A memory inspection panel is currently in development.
+
+**Q: Can I delete a memory?**
+The ability to query, correct, and delete specific memories is coming soon.
+
+**Q: Is my data private?**
+Yes — by default everything is stored locally on your machine using LanceDB.
+
+---
+
+## 🚀 Coming Soon
+
+AlphaAvatar's Memory plugin is actively being improved. Here's what's coming:
+
+| Feature | Description |
+|---------|-------------|
+| 🎤 Voice Memory | Remember things said during voice conversations |
+| 👁️ Visual Memory | Extract memories from images, screenshots, and camera frames |
+| 😊 Face-based Memory | Remember users by their face using facial recognition |
+| 🕸️ Graph Memory | Understand relationships between memories (people, events, goals) |
+| 🔒 Memory Privacy Controls | Let users view, edit, export, or delete their memories |
+| 👥 Multi-user Memory | Separate memories for multiple users in the same session |
+
+---
+
+## 📚 Related Links
+
+- [Qdrant Documentation](https://qdrant.tech)
+- [LangChain Documentation](https://www.langchain.com)
+- [AlphaAvatar ROADMAP — Memory Section](https://github.com/AlphaAvatar/AlphaAvatar/blob/main/ROADMAP.md#-memory)
+- [Persona Plugin](https://github.com/AlphaAvatar/AlphaAvatar/blob/main/avatar-plugins/avatar-plugins-persona/README.md)
+- 

alpha_avatar_plugins_memory-0.6.2/README.md

@@ -0,0 +1,149 @@
+# 🧠 Memory Plugin for AlphaAvatar
+
+> Give AlphaAvatar the ability to **remember you** — across conversations, sessions, and time.
+
+---
+
+## 🤔 What is the Memory Plugin?
+
+Imagine talking to an assistant that forgets everything you said the moment you close the app. Frustrating, right?
+
+The **Memory Plugin** solves this. It allows AlphaAvatar to:
+- Remember your **name, preferences, and habits**
+- Recall **past conversations** and what was discussed
+- Learn from its own **tool usage** (like web searches or documents it read)
+- Get **smarter and more personal** the more you use it
+
+Think of it like giving AlphaAvatar a notebook — every important thing gets written down and read back when needed.
+
+---
+
+## 💡 How Does it Work? (Simple Flow)
+
+```
+You say something to AlphaAvatar
+            ↓
+AlphaAvatar picks out important information
+(your name, preferences, facts, decisions)
+            ↓
+That information gets saved to local storage
+            ↓
+Next time you start a conversation
+            ↓
+AlphaAvatar reads its saved memories
+            ↓
+Responds like it knows you — because it does
+```
+
+No setup needed from your side. It works automatically in the background.
+
+---
+
+## ✨ Features
+
+### 🌍 Global Memory (Avatar-wide)
+AlphaAvatar maintains memories across your entire experience — not just one chat window.
+
+This includes:
+- Things **you** told it (name, preferences, goals)
+- Things it learned from **external tools** (web searches, documents, research)
+- Things it observed from **its own environment** (what worked, what didn't)
+
+### ⚡ Real-time Context Updates
+Memory doesn't just update at the end of a conversation — it updates **while you're talking**. If you mention something new mid-conversation, it gets remembered immediately.
+
+---
+
+## 📦 What Kind of Things Does AlphaAvatar Remember?
+
+| Type | Real Example |
+|------|-------------|
+| 👤 Personal facts | "My name is Alex and I live in New York" |
+| ❤️ Preferences | "I like short, direct answers" |
+| 🔍 Research results | Facts found during a DeepResearch session |
+| 📄 Document knowledge | Content from files you uploaded via RAG |
+| 🗓️ Past conversations | Topics and decisions from previous sessions |
+| 🛠️ Tool interactions | Results from MCP tools like Notion or Gmail |
+
+---
+
+## 🔧 Installation
+
+```bash
+pip install alpha-avatar-plugins-memory
+```
+
+That's it. The plugin is automatically used when you run AlphaAvatar.
+
+---
+
+## 🗄️ How Memories are Stored
+
+AlphaAvatar uses **vector storage** to save memories. This means memories aren't stored as plain text — they're stored in a way that makes them easy to search and retrieve intelligently.
+
+### Supported Backends
+
+| Backend | What it Does |
+|---------|-------------|
+| **LanceDB** | Saves memories locally on your machine (default) |
+| **Qdrant** | A powerful vector database for storage and retrieval |
+| **LangChain** | The pipeline that extracts and processes memories |
+
+By default, everything is stored **locally** — your memories never leave your machine unless you configure a cloud backend.
+
+---
+
+## 🔗 How Memory Connects to Other Plugins
+
+The Memory plugin doesn't work alone — it's the foundation that other plugins build on:
+
+```
+Memory Plugin
+      ├── → Persona Plugin   (builds your user profile from memories)
+      ├── → RAG Plugin       (stores document knowledge in memory)
+      ├── → Reflection*      (analyzes memories to improve Avatar behavior)
+      └── → Planning*        (uses memories to set reminders and goals)
+
+* Coming soon
+```
+
+---
+
+## 🙋 Common Questions
+
+**Q: Does memory work across different devices?**
+By default, memory is stored locally. Cross-device sync is not yet supported.
+
+**Q: Can I see what AlphaAvatar remembers about me?**
+A memory inspection panel is currently in development.
+
+**Q: Can I delete a memory?**
+The ability to query, correct, and delete specific memories is coming soon.
+
+**Q: Is my data private?**
+Yes — by default everything is stored locally on your machine using LanceDB.
+
+---
+
+## 🚀 Coming Soon
+
+AlphaAvatar's Memory plugin is actively being improved. Here's what's coming:
+
+| Feature | Description |
+|---------|-------------|
+| 🎤 Voice Memory | Remember things said during voice conversations |
+| 👁️ Visual Memory | Extract memories from images, screenshots, and camera frames |
+| 😊 Face-based Memory | Remember users by their face using facial recognition |
+| 🕸️ Graph Memory | Understand relationships between memories (people, events, goals) |
+| 🔒 Memory Privacy Controls | Let users view, edit, export, or delete their memories |
+| 👥 Multi-user Memory | Separate memories for multiple users in the same session |
+
+---
+
+## 📚 Related Links
+
+- [Qdrant Documentation](https://qdrant.tech)
+- [LangChain Documentation](https://www.langchain.com)
+- [AlphaAvatar ROADMAP — Memory Section](https://github.com/AlphaAvatar/AlphaAvatar/blob/main/ROADMAP.md#-memory)
+- [Persona Plugin](https://github.com/AlphaAvatar/AlphaAvatar/blob/main/avatar-plugins/avatar-plugins-persona/README.md)
+- 

{alpha_avatar_plugins_memory-0.6.0 → alpha_avatar_plugins_memory-0.6.2}/alphaavatar/plugins/memory/__init__.py

@@ -14,10 +14,10 @@
 import os
 
 from alphaavatar.agents import AvatarModule, AvatarPlugin
-from alphaavatar.agents.utils.files.work_dirs import UserPath
+from alphaavatar.agents.runtime import SessionRuntime
 
 from .log import logger
-from .memory_langchain import MemoryLangchain
+from .memory_runtime import MemoryRuntime
 from .version import __version__
 
 __all__ = [
@@ -25,7 +25,7 @@ __all__ = [
 ]
 
 
-class MemoryLangchainPlugin(AvatarPlugin):
+class MemoryPlugin(AvatarPlugin):
     def __init__(self) -> None:
         super().__init__(__name__, __version__, __package__, logger)  # type: ignore
 
@@ -33,24 +33,24 @@ class MemoryLangchainPlugin(AvatarPlugin):
 
     def get_plugin(
         self,
-        user_path: UserPath,
+        session_runtime: SessionRuntime,
         memory_search_context: int,
         memory_recall_num: int,
         maximum_memory_num: int,
         memory_init_config: dict,
         *args,
         **kwargs,
-    ) -> MemoryLangchain:
+    ) -> MemoryRuntime:
         try:
-            return MemoryLangchain(
-                user_path=user_path,
+            return MemoryRuntime(
+                session_runtime=session_runtime,
                 memory_search_context=memory_search_context,
                 memory_recall_num=memory_recall_num,
                 maximum_memory_num=maximum_memory_num,
-                memory_init_config=memory_init_config,
+                **memory_init_config,
             )
         except Exception as e:
-            raise ImportError(f"Failed to initialize MemoryLangchain plugin: {e}") from e
+            raise ImportError(f"Failed to initialize MemoryRuntime plugin: {e}") from e
 
 
 def bootstrap_inference_runners() -> None:
@@ -82,7 +82,7 @@ def bootstrap_inference_runners() -> None:
 AvatarPlugin.register_avatar_plugin(
     AvatarModule.MEMORY,
     "default",
-    MemoryLangchainPlugin(),
+    MemoryPlugin(),
 )
 
 # Runner bootstrap register

alpha_avatar_plugins_memory-0.6.0/alphaavatar/plugins/memory/memory_langchain.py → alpha_avatar_plugins_memory-0.6.2/alphaavatar/plugins/memory/memory_runtime.py

@@ -19,7 +19,6 @@ import re
 from typing import Any
 
 from langchain_core.prompts import ChatPromptTemplate
-from langchain_openai import ChatOpenAI
 from livekit.agents.job import get_job_context
 from livekit.agents.llm import ChatItem
 from pydantic import BaseModel, Field
@@ -32,8 +31,11 @@ from alphaavatar.agents.memory import (
     MemoryType,
     VectorRunnerOP,
 )
-from alphaavatar.agents.utils import format_current_time
-from alphaavatar.agents.utils.files.work_dirs import UserPath
+from alphaavatar.agents.providers import (
+    ProviderGateway,
+    ProvidersConfig,
+)
+from alphaavatar.agents.runtime import SessionRuntime
 
 from .log import logger
 from .memory_markdown import save_memory_items_to_markdown
@@ -309,45 +311,46 @@ def _select_by_priority(
     return picked
 
 
-class MemoryInitConfig(BaseModel):
-    openai_model: str = Field(default="gpt-4o-mini")
-    temperature: float = Field(default=0.0)
+class MemoryProviderConfig(BaseModel):
+    conversation_delta_task: str = "memory.conversation_delta"
+    tool_delta_task: str = "memory.tool_delta"
+    gateway: ProvidersConfig = Field(default_factory=ProvidersConfig)
 
 
-class MemoryLangchain(MemoryBase):
+class MemoryRuntime(MemoryBase):
     def __init__(
         self,
         *,
-        user_path: UserPath,
+        session_runtime: SessionRuntime,
         memory_search_context: int = 3,
         memory_recall_num: int = 10,
         maximum_memory_num: int = 24,
-        memory_init_config: dict[str, Any] | None = None,
+        provider: dict[str, Any] | None = None,
+        **kwargs,
     ) -> None:
         super().__init__(
-            user_path=user_path,
+            session_runtime=session_runtime,
             memory_search_context=memory_search_context,
             memory_recall_num=memory_recall_num,
             maximum_memory_num=maximum_memory_num,
         )
 
-        self._memory_init_config = (
-            MemoryInitConfig(**memory_init_config) if memory_init_config else MemoryInitConfig()
+        self._provider_config = (
+            MemoryProviderConfig(**provider) if provider else MemoryProviderConfig()
         )
 
-        llm = ChatOpenAI(
-            model=self._memory_init_config.openai_model,
-            temperature=self._memory_init_config.temperature,
-        )  # type: ignore
+        self._conversation_delta_task = self._provider_config.conversation_delta_task
+        self._tool_delta_task = self._provider_config.tool_delta_task
 
-        self._delta_llm = llm.with_structured_output(MemoryDelta)
-        self._conversation_delta_chain = CONVERSATION_DELTA_PROMPT | self._delta_llm
-        self._tool_delta_chain = TOOL_DELTA_PROMPT | self._delta_llm
-        self._executor = get_job_context().inference_executor
+        self._provider_gateway = ProviderGateway(self._provider_config.gateway)
+        self._provider_gateway.validate_tasks(
+            [
+                self._conversation_delta_task,
+                self._tool_delta_task,
+            ]
+        )
 
-    @property
-    def memory_init_config(self) -> MemoryInitConfig:
-        return self._memory_init_config
+        self._executor = get_job_context().inference_executor
 
     @property
     def inference_method(self) -> str:
@@ -360,45 +363,94 @@ class MemoryLangchain(MemoryBase):
             )
         return method
 
+    async def _safe_ainvoke_delta(
+        self,
+        *,
+        task_name: str,
+        prompt,
+        payload: dict[str, Any],
+        metadata: dict[str, Any],
+        timeout: float = 12.0,
+    ) -> MemoryDelta:
+        try:
+            result = await asyncio.wait_for(
+                self._provider_gateway.ainvoke_structured(
+                    task_name=task_name,
+                    prompt=prompt,
+                    payload=payload,
+                    output_schema=MemoryDelta,
+                    metadata=metadata,
+                ),
+                timeout=timeout,
+            )
+
+            if isinstance(result.output, MemoryDelta):
+                return result.output
+
+            return MemoryDelta.model_validate(result.output)
+
+        except asyncio.TimeoutError:
+            logger.warning("[Memory] delta extraction timeout task=%s", task_name)
+            return MemoryDelta()
+        except Exception:
+            logger.exception("[Memory] delta extraction failed task=%s", task_name)
+            return MemoryDelta()
+
     async def _safe_ainvoke_conversation_delta(
         self,
         *,
         message_content: str,
+        memory_cache: MemoryCache,
         timeout: float = 12.0,
     ) -> MemoryDelta:
         payload = {
             "type": MemoryType.CONVERSATION,
             "message_content": message_content,
         }
-        try:
-            return await asyncio.wait_for(
-                self._conversation_delta_chain.ainvoke(payload), timeout=timeout
-            )  # type: ignore
-        except asyncio.TimeoutError:
-            logger.warning("[Memory] conversation delta extraction timeout")
-            return MemoryDelta()
-        except Exception:
-            logger.exception("[Memory] conversation delta extraction failed")
-            return MemoryDelta()
+
+        return await self._safe_ainvoke_delta(
+            task_name=self._conversation_delta_task,
+            prompt=CONVERSATION_DELTA_PROMPT,
+            payload=payload,
+            metadata={
+                "provider_dir": memory_cache.provider_dir,
+                "plugin": "memory",
+                "component": "memory_delta_extractor",
+                "operation": "conversation_delta",
+                "session_id": memory_cache.session_id,
+                "user_or_tool_id": memory_cache.user_or_tool_id,
+                "memory_type": str(memory_cache.type),
+            },
+            timeout=timeout,
+        )
 
     async def _safe_ainvoke_tool_delta(
         self,
         *,
         message_content: str,
+        memory_cache: MemoryCache,
         timeout: float = 12.0,
     ) -> MemoryDelta:
         payload = {
             "type": MemoryType.TOOLS,
             "message_content": message_content,
         }
-        try:
-            return await asyncio.wait_for(self._tool_delta_chain.ainvoke(payload), timeout=timeout)  # type: ignore
-        except asyncio.TimeoutError:
-            logger.warning("[Memory] tool delta extraction timeout")
-            return MemoryDelta()
-        except Exception:
-            logger.exception("[Memory] tool delta extraction failed")
-            return MemoryDelta()
+
+        return await self._safe_ainvoke_delta(
+            task_name=self._tool_delta_task,
+            prompt=TOOL_DELTA_PROMPT,
+            payload=payload,
+            metadata={
+                "provider_dir": memory_cache.provider_dir,
+                "plugin": "memory",
+                "component": "memory_delta_extractor",
+                "operation": "tool_delta",
+                "session_id": memory_cache.session_id,
+                "user_or_tool_id": memory_cache.user_or_tool_id,
+                "memory_type": str(memory_cache.type),
+            },
+            timeout=timeout,
+        )
 
     def _apply_delta_to_bucket(
         self,
@@ -408,7 +460,7 @@ class MemoryLangchain(MemoryBase):
         memory_cache: MemoryCache,
         user_or_tool_memory_type: MemoryType,
     ):
-        updated_time = format_current_time().time_str
+        updated_time = memory_cache.time
         assistant_memories: list[MemoryItem] = []
         target_memories: list[MemoryItem] = []
 
@@ -563,16 +615,19 @@ class MemoryLangchain(MemoryBase):
                     conversation_delta, tool_delta = await asyncio.gather(
                         self._safe_ainvoke_conversation_delta(
                             message_content=message_content,
+                            memory_cache=cache,
                             timeout=30.0,
                         ),
                         self._safe_ainvoke_tool_delta(
                             message_content=message_content,
+                            memory_cache=cache,
                             timeout=30.0,
                         ),
                     )
                 else:
                     conversation_delta = await self._safe_ainvoke_conversation_delta(
                         message_content=message_content,
+                        memory_cache=cache,
                         timeout=30.0,
                     )
                     tool_delta = None
@@ -664,8 +719,8 @@ class MemoryLangchain(MemoryBase):
         ## 5.1 Save to local .md file for backup/debug
         try:
             md_result = save_memory_items_to_markdown(
-                avatar_memory_path=self._avatar_memory_path,
-                session_memory_path=self.working_dir,
+                avatar_memory_path=self.session_runtime.avatar_path.memory_dir,
+                session_memory_path=self.session_runtime.session_path.memory_dir,
                 memory_items=memory_items,
             )
             logger.info(f"Memory local markdown backup success: {md_result}")

{alpha_avatar_plugins_memory-0.6.0 → alpha_avatar_plugins_memory-0.6.2}/alphaavatar/plugins/memory/runner/lancedb_runner.py

@@ -13,11 +13,14 @@
 # limitations under the License.
 import json
 import os
+from typing import Any
 
 from livekit.agents.inference_runner import _InferenceRunner
 
 from alphaavatar.agents.memory import VectorRunnerOP
-from alphaavatar.agents.utils.vdb import embedding, lancedb
+from alphaavatar.agents.providers import ProviderKind, ProviderTaskConfig
+from alphaavatar.agents.providers.embedding import create_embedding_model
+from alphaavatar.agents.utils.vdb import lancedb
 
 
 class LanceDBRunner(_InferenceRunner):
@@ -174,6 +177,39 @@ class LanceDBRunner(_InferenceRunner):
 
         return result
 
+    #
+    # Runner Interface
+    #
+
+    def _get_vdb_config(self, config: dict[str, Any]) -> dict[str, Any]:
+        vdb_config = dict(config)
+        vdb_config.pop("embedding", None)
+        return vdb_config
+
+    def _get_memory_embeddings(self, config: dict[str, Any]):
+        embedding_config = config.get("embedding")
+
+        if not embedding_config:
+            raise ValueError("`embedding` is required in MEMORY_VDB_CONFIG")
+
+        provider = embedding_config.get("provider")
+        model = embedding_config.get("model")
+        extra = embedding_config.get("extra") or {}
+
+        if not provider:
+            raise ValueError("`embedding.provider` is required in MEMORY_VDB_CONFIG")
+        if not model:
+            raise ValueError("`embedding.model` is required in MEMORY_VDB_CONFIG")
+
+        task_config = ProviderTaskConfig(
+            kind=ProviderKind.EMBEDDING,
+            provider=provider,
+            model=model,
+            extra=extra,
+        )
+
+        return create_embedding_model(task_config)
+
     def initialize(self) -> None:
         config = os.getenv("MEMORY_VDB_CONFIG", "{}")
         config = json.loads(config)
@@ -182,9 +218,9 @@ class LanceDBRunner(_InferenceRunner):
         if not self._collection_name:
             raise ValueError("collection_name is required in MEMORY_VDB_CONFIG")
 
-        self._client = lancedb.get_client(**config)
+        self._client = lancedb.get_client(**self._get_vdb_config(config))
 
-        self._embeddings = embedding.get_model(**config)
+        self._embeddings = self._get_memory_embeddings(config)
         embedding_dim = len(self._embeddings.embed_query("dimension-probe"))
 
         self._ensure_collection(self._collection_name, embedding_dim)

{alpha_avatar_plugins_memory-0.6.0 → alpha_avatar_plugins_memory-0.6.2}/alphaavatar/plugins/memory/runner/qdrant_runner.py

@@ -13,6 +13,7 @@
 # limitations under the License.
 import json
 import os
+from typing import Any
 
 from langchain_qdrant import QdrantVectorStore
 from livekit.agents.inference_runner import _InferenceRunner
@@ -27,7 +28,9 @@ from qdrant_client.models import (
 )
 
 from alphaavatar.agents.memory import VectorRunnerOP
-from alphaavatar.agents.utils.vdb import embedding, qdrant
+from alphaavatar.agents.providers import ProviderKind, ProviderTaskConfig
+from alphaavatar.agents.providers.embedding import create_embedding_model
+from alphaavatar.agents.utils.vdb import qdrant
 
 
 class QdrantRunner(_InferenceRunner):
@@ -127,6 +130,39 @@ class QdrantRunner(_InferenceRunner):
 
         return result
 
+    #
+    # Runner Interface
+    #
+
+    def _get_vdb_config(self, config: dict[str, Any]) -> dict[str, Any]:
+        vdb_config = dict(config)
+        vdb_config.pop("embedding", None)
+        return vdb_config
+
+    def _get_memory_embeddings(self, config: dict[str, Any]):
+        embedding_config = config.get("embedding")
+
+        if not embedding_config:
+            raise ValueError("`embedding` is required in MEMORY_VDB_CONFIG")
+
+        provider = embedding_config.get("provider")
+        model = embedding_config.get("model")
+        extra = embedding_config.get("extra") or {}
+
+        if not provider:
+            raise ValueError("`embedding.provider` is required in MEMORY_VDB_CONFIG")
+        if not model:
+            raise ValueError("`embedding.model` is required in MEMORY_VDB_CONFIG")
+
+        task_config = ProviderTaskConfig(
+            kind=ProviderKind.EMBEDDING,
+            provider=provider,
+            model=model,
+            extra=extra,
+        )
+
+        return create_embedding_model(task_config)
+
     def initialize(self) -> None:
         # get config
         config = os.getenv("MEMORY_VDB_CONFIG", "{}")
@@ -134,10 +170,10 @@ class QdrantRunner(_InferenceRunner):
         self._collection_name = config.get("collection_name", None)
 
         # init client
-        self._client = qdrant.get_client(**config)
+        self._client = qdrant.get_client(**self._get_vdb_config(config))
 
         # init memory
-        self._embeddings = embedding.get_model(**config)
+        self._embeddings = self._get_memory_embeddings(config)
         self._ensure_collection(
             self._collection_name,
             len(self._embeddings.embed_query("dimension-probe")),

{alpha_avatar_plugins_memory-0.6.0 → alpha_avatar_plugins_memory-0.6.2}/alphaavatar/plugins/memory/version.py

@@ -11,4 +11,4 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-__version__ = "0.6.0"
+__version__ = "0.6.2"

{alpha_avatar_plugins_memory-0.6.0 → alpha_avatar_plugins_memory-0.6.2}/pyproject.toml

@@ -8,7 +8,7 @@ dynamic = ["version"]
 description = "AlphaAvatar Framework plugin for memory service"
 readme = "README.md"
 license = "Apache-2.0"
-requires-python = ">=3.10"
+requires-python = ">=3.11,<3.12"
 keywords = ["llm", "agents", "AI", "memory", "langchain"]
 classifiers = [
     "Intended Audience :: Developers",
@@ -26,9 +26,8 @@ classifiers = [
 dependencies = [
     "langchain-core",
     "langchain-community",
-    "langchain-openai",
     "qdrant-client",
-    "lancedb",
+    "lancedb==0.30.2",
     "langchain-qdrant",
     "torchaudio>=0.10.1"
 ]

alpha_avatar_plugins_memory-0.6.0/PKG-INFO

@@ -1,54 +0,0 @@
-Metadata-Version: 2.4
-Name: alpha-avatar-plugins-memory
-Version: 0.6.0
-Summary: AlphaAvatar Framework plugin for memory service
-Project-URL: Source, https://github.com/AlphaAvatar/AlphaAvatar
-License-Expression: Apache-2.0
-Keywords: AI,agents,langchain,llm,memory
-Classifier: Intended Audience :: Developers
-Classifier: License :: OSI Approved :: Apache Software License
-Classifier: Programming Language :: Python :: 3
-Classifier: Programming Language :: Python :: 3 :: Only
-Classifier: Programming Language :: Python :: 3.10
-Classifier: Programming Language :: Python :: 3.11
-Classifier: Programming Language :: Python :: 3.12
-Classifier: Programming Language :: Python :: 3.13
-Classifier: Topic :: Multimedia :: Sound/Audio
-Classifier: Topic :: Multimedia :: Video
-Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
-Requires-Python: >=3.10
-Requires-Dist: lancedb
-Requires-Dist: langchain-community
-Requires-Dist: langchain-core
-Requires-Dist: langchain-openai
-Requires-Dist: langchain-qdrant
-Requires-Dist: qdrant-client
-Requires-Dist: torchaudio>=0.10.1
-Description-Content-Type: text/markdown
-
-# Memory Plugin for AlphaAvatar
-
-Support for storing and managing conversational memory within Avatar, with a unified interface to open-source memory frameworks.
-This plugin simplifies memory management so developers don’t need to worry about storage limits, or backend details.
-
-## Features
-
-* **Global Memory (Avatar-wide):** Retrieval and updates of cross-user, Avatar-level memories. This includes memories derived from automatically identified users as well as memories created through the Avatar’s interactions with external tools and the environment.
-* **Real-time Context Updates:** Memories are written and refreshed on the fly based on the current dialogue context.
-
-## Installation
-
-```bash
-pip install alpha-avatar-plugins-memory
-```
-
----
-
-## Supported Open-Source Memory Frameworks
-
-### Default: **LangChain + Qdrant**
-
-| Module                 | Description                                                                                   | Docs                                                                   |
-| ---------------------- | --------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------- |
-| **Qdrant** (default) | Combined framework for persona extraction pipelines and vector storage / retrieval.           | [Qdrant](https://qdrant.tech) |
-| **LangChain** (default) | Combined framework for persona extraction pipelines and vector storage / retrieval.           | [LangChain](https://www.langchain.com) |

alpha_avatar_plugins_memory-0.6.0/README.md

@@ -1,26 +0,0 @@
-# Memory Plugin for AlphaAvatar
-
-Support for storing and managing conversational memory within Avatar, with a unified interface to open-source memory frameworks.
-This plugin simplifies memory management so developers don’t need to worry about storage limits, or backend details.
-
-## Features
-
-* **Global Memory (Avatar-wide):** Retrieval and updates of cross-user, Avatar-level memories. This includes memories derived from automatically identified users as well as memories created through the Avatar’s interactions with external tools and the environment.
-* **Real-time Context Updates:** Memories are written and refreshed on the fly based on the current dialogue context.
-
-## Installation
-
-```bash
-pip install alpha-avatar-plugins-memory
-```
-
----
-
-## Supported Open-Source Memory Frameworks
-
-### Default: **LangChain + Qdrant**
-
-| Module                 | Description                                                                                   | Docs                                                                   |
-| ---------------------- | --------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------- |
-| **Qdrant** (default) | Combined framework for persona extraction pipelines and vector storage / retrieval.           | [Qdrant](https://qdrant.tech) |
-| **LangChain** (default) | Combined framework for persona extraction pipelines and vector storage / retrieval.           | [LangChain](https://www.langchain.com) |