cognee 0.5.4.dev1__tar.gz → 0.5.5.dev0__tar.gz

{cognee-0.5.4.dev1 → cognee-0.5.5.dev0}/.env.template

@@ -20,12 +20,19 @@ LLM_MODEL="openai/gpt-5-mini"
 LLM_PROVIDER="openai"
 LLM_ENDPOINT=""
 LLM_API_VERSION=""
-LLM_MAX_TOKENS="16384"
+# Cognee uses this value to determine optimal chunk size. (This value is only used by Cognee to determine chunk size and is not forwarded in LLM calls)
+# LLM_MAX_TOKENS="16384"
 # Instructor's modes determine how structured data is requested from and extracted from LLM responses
 # You can change this type (i.e. mode) via this env variable
 # Each LLM has its own default value, e.g. gpt-5 models have "json_schema_mode"
 LLM_INSTRUCTOR_MODE=""
 
+# Extra kwargs passed to every LLM completion call (JSON string).
+# These are forwarded as-is to the underlying litellm/instructor API.
+# Examples:
+#   LLM_ARGS='{"max_tokens": 16384, "temperature": 0.7, "top_p": 0.9}'
+# LLM_ARGS='{}'
+
 EMBEDDING_PROVIDER="openai"
 EMBEDDING_MODEL="openai/text-embedding-3-large"
 EMBEDDING_ENDPOINT=""

{cognee-0.5.4.dev1 → cognee-0.5.5.dev0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cognee
-Version: 0.5.4.dev1
+Version: 0.5.5.dev0
 Summary: Cognee - is a library for enriching LLM context with a semantic layer for better understanding and reasoning.
 Project-URL: Homepage, https://www.cognee.ai
 Project-URL: Repository, https://github.com/topoteretes/cognee
@@ -37,7 +37,6 @@ Requires-Dist: lancedb<1.0.0,>=0.24.0
 Requires-Dist: langdetect>=1.0.9
 Requires-Dist: limits<5,>=4.4.1
 Requires-Dist: litellm>=1.76.0
-Requires-Dist: mistralai>=1.9.10
 Requires-Dist: nbformat<6.0.0,>=5.7.0
 Requires-Dist: networkx<4,>=3.4.2
 Requires-Dist: numpy<=4.0.0,>=1.26.4
@@ -128,6 +127,7 @@ Provides-Extra: llama-index
 Requires-Dist: llama-index-core<0.14,>=0.13.0; extra == 'llama-index'
 Provides-Extra: mistral
 Requires-Dist: mistral-common<2,>=1.5.2; extra == 'mistral'
+Requires-Dist: mistralai<2,>=1.9.10; extra == 'mistral'
 Provides-Extra: monitoring
 Requires-Dist: langfuse<3,>=2.32.0; extra == 'monitoring'
 Requires-Dist: opentelemetry-api<2,>=1.20.0; extra == 'monitoring'
@@ -180,7 +180,7 @@ Description-Content-Type: text/markdown
   Cognee - Build AI memory with a Knowledge Engine that learns
 
   <p align="center">
-  <a href="https://www.youtube.com/watch?v=1bezuvLwJmw&t=2s">Demo</a>
+  <a href="https://www.youtube.com/watch?v=8hmqS2Y5RVQ&t=13s">Demo</a>
   .
   <a href="https://docs.cognee.ai/">Docs</a>
   .
@@ -204,10 +204,6 @@ Description-Content-Type: text/markdown
   <a href="https://github.com/sponsors/topoteretes"><img src="https://img.shields.io/badge/Sponsor-❤️-ff69b4.svg" alt="Sponsor"></a>
 
 <p>
-  <a href="https://www.producthunt.com/posts/cognee?embed=true&utm_source=badge-top-post-badge&utm_medium=badge&utm_souce=badge-cognee" target="_blank" style="display:inline-block; margin-right:10px;">
-    <img src="https://api.producthunt.com/widgets/embed-image/v1/top-post-badge.svg?post_id=946346&theme=light&period=daily&t=1744472480704" alt="cognee - Memory&#0032;for&#0032;AI&#0032;Agents&#0032;&#0032;in&#0032;5&#0032;lines&#0032;of&#0032;code | Product Hunt" width="250" height="54" />
-  </a>
-
   <a href="https://trendshift.io/repositories/13955" target="_blank" style="display:inline-block;">
     <img src="https://trendshift.io/api/badge/repositories/13955" alt="topoteretes%2Fcognee | Trendshift" width="250" height="55" />
   </a>
@@ -231,7 +227,7 @@ Use our knowledge engine to build personalized and dynamic memory for AI Agents.
 
 
 <div style="text-align: center">
-  <img src="https://raw.githubusercontent.com/topoteretes/cognee/refs/heads/main/assets/cognee_benefits.png" alt="Why cognee?" width="50%" />
+  <img src="https://raw.githubusercontent.com/topoteretes/cognee/refs/heads/main/assets/cognee_benefits.png" alt="Why cognee?" width="80%" />
 </div>
 </div>
 
@@ -240,21 +236,19 @@ Use our knowledge engine to build personalized and dynamic memory for AI Agents.
 
 ## About Cognee
 
-Cognee is an open-source knowledge engine that transforms your raw data into persistent and dynamic AI memory for Agents. It combines vector search, graph databases and self-improvement to make your documents both searchable by meaning and connected by relationships as they change and evolve.
+Cognee is an open-source knowledge engine that lets you ingest data in any format or structure and continuously learns to provide the right context for AI agents. It combines vector search, graph databases and cognitive science approaches to make your documents both searchable by meaning and connected by relationships as they change and evolve.
 
-Cognee offers default knowledge creation and search which we describe bellow. But with Cognee you can build your modular knowledge blocks!
 
 
 :star: _Help us reach more developers and grow the cognee community. Star this repo!_
 
 
-### Cognee Open Source:
+### Why use Cognee:
+
+- Knowledge infrastructure — unified ingestion, graph/vector search, runs locally, ontology grounding, multimodal
+- Persistent and Learning Agents - learn from feedback, context management, cross-agent knowledge sharing
+- Reliable and Trustworthy Agents - agentic user/tenant isolation, traceability, OTEL collector, audit traits
 
-- Interconnects any type of data — including past conversations, files, images, and audio transcriptions
-- Replaces traditional database lookups with a unified knowledge engine built with graphs and vectors
-- Reduces developer effort and infrastructure cost while improving quality and precision
-- Provides Pythonic data pipelines for ingestion from 30+ data sources
-- Offers high customizability through user-defined tasks, modular pipelines, and built-in search endpoints
 
 
 ## Basic Usage & Feature Guide
@@ -290,7 +284,7 @@ To integrate other LLM providers, see our [LLM Provider Documentation](https://d
 
 ### Step 3: Run the Pipeline
 
-Cognee will take your documents, generate a knowledge graph from them and then query the graph based on combined relationships.
+Cognee will take your documents, load them into the knowledge angine and search combined vector/graph relationships.
 
 Now, run a minimal pipeline:
 
@@ -304,12 +298,9 @@ async def main():
     # Add text to cognee
     await cognee.add("Cognee turns documents into AI memory.")
 
-    # Generate the knowledge graph
+    # Add to knowledge engine
     await cognee.cognify()
 
-    # Add memory algorithms to the graph
-    await cognee.memify()
-
     # Query the knowledge graph
     results = await cognee.search("What does Cognee do?")
 
@@ -354,15 +345,7 @@ See Cognee in action:
 
 ### Persistent Agent Memory
 
-[Cognee Memory for LangGraph Agents](https://github.com/user-attachments/assets/e113b628-7212-4a2b-b288-0be39a93a1c3)
-
-### Simple GraphRAG
-
-[Watch Demo](https://github.com/user-attachments/assets/f2186b2e-305a-42b0-9c2d-9f4473f15df8)
-
-### Cognee with Ollama
-
-[Watch Demo](https://github.com/user-attachments/assets/39672858-f774-4136-b957-1e2de67b8981)
+[![Watch Demo](https://img.youtube.com/vi/8hmqS2Y5RVQ/maxresdefault.jpg)](https://www.youtube.com/watch?v=8hmqS2Y5RVQ&t=13s)
 
 
 ## Community & Support

{cognee-0.5.4.dev1 → cognee-0.5.5.dev0}/README.md

@@ -8,7 +8,7 @@
   Cognee - Build AI memory with a Knowledge Engine that learns
 
   <p align="center">
-  <a href="https://www.youtube.com/watch?v=1bezuvLwJmw&t=2s">Demo</a>
+  <a href="https://www.youtube.com/watch?v=8hmqS2Y5RVQ&t=13s">Demo</a>
   .
   <a href="https://docs.cognee.ai/">Docs</a>
   .
@@ -32,10 +32,6 @@
   <a href="https://github.com/sponsors/topoteretes"><img src="https://img.shields.io/badge/Sponsor-❤️-ff69b4.svg" alt="Sponsor"></a>
 
 <p>
-  <a href="https://www.producthunt.com/posts/cognee?embed=true&utm_source=badge-top-post-badge&utm_medium=badge&utm_souce=badge-cognee" target="_blank" style="display:inline-block; margin-right:10px;">
-    <img src="https://api.producthunt.com/widgets/embed-image/v1/top-post-badge.svg?post_id=946346&theme=light&period=daily&t=1744472480704" alt="cognee - Memory&#0032;for&#0032;AI&#0032;Agents&#0032;&#0032;in&#0032;5&#0032;lines&#0032;of&#0032;code | Product Hunt" width="250" height="54" />
-  </a>
-
   <a href="https://trendshift.io/repositories/13955" target="_blank" style="display:inline-block;">
     <img src="https://trendshift.io/api/badge/repositories/13955" alt="topoteretes%2Fcognee | Trendshift" width="250" height="55" />
   </a>
@@ -59,7 +55,7 @@ Use our knowledge engine to build personalized and dynamic memory for AI Agents.
 
 
 <div style="text-align: center">
-  <img src="https://raw.githubusercontent.com/topoteretes/cognee/refs/heads/main/assets/cognee_benefits.png" alt="Why cognee?" width="50%" />
+  <img src="https://raw.githubusercontent.com/topoteretes/cognee/refs/heads/main/assets/cognee_benefits.png" alt="Why cognee?" width="80%" />
 </div>
 </div>
 
@@ -68,21 +64,19 @@ Use our knowledge engine to build personalized and dynamic memory for AI Agents.
 
 ## About Cognee
 
-Cognee is an open-source knowledge engine that transforms your raw data into persistent and dynamic AI memory for Agents. It combines vector search, graph databases and self-improvement to make your documents both searchable by meaning and connected by relationships as they change and evolve.
+Cognee is an open-source knowledge engine that lets you ingest data in any format or structure and continuously learns to provide the right context for AI agents. It combines vector search, graph databases and cognitive science approaches to make your documents both searchable by meaning and connected by relationships as they change and evolve.
 
-Cognee offers default knowledge creation and search which we describe bellow. But with Cognee you can build your modular knowledge blocks!
 
 
 :star: _Help us reach more developers and grow the cognee community. Star this repo!_
 
 
-### Cognee Open Source:
+### Why use Cognee:
+
+- Knowledge infrastructure — unified ingestion, graph/vector search, runs locally, ontology grounding, multimodal
+- Persistent and Learning Agents - learn from feedback, context management, cross-agent knowledge sharing
+- Reliable and Trustworthy Agents - agentic user/tenant isolation, traceability, OTEL collector, audit traits
 
-- Interconnects any type of data — including past conversations, files, images, and audio transcriptions
-- Replaces traditional database lookups with a unified knowledge engine built with graphs and vectors
-- Reduces developer effort and infrastructure cost while improving quality and precision
-- Provides Pythonic data pipelines for ingestion from 30+ data sources
-- Offers high customizability through user-defined tasks, modular pipelines, and built-in search endpoints
 
 
 ## Basic Usage & Feature Guide
@@ -118,7 +112,7 @@ To integrate other LLM providers, see our [LLM Provider Documentation](https://d
 
 ### Step 3: Run the Pipeline
 
-Cognee will take your documents, generate a knowledge graph from them and then query the graph based on combined relationships.
+Cognee will take your documents, load them into the knowledge angine and search combined vector/graph relationships.
 
 Now, run a minimal pipeline:
 
@@ -132,12 +126,9 @@ async def main():
     # Add text to cognee
     await cognee.add("Cognee turns documents into AI memory.")
 
-    # Generate the knowledge graph
+    # Add to knowledge engine
     await cognee.cognify()
 
-    # Add memory algorithms to the graph
-    await cognee.memify()
-
     # Query the knowledge graph
     results = await cognee.search("What does Cognee do?")
 
@@ -182,15 +173,7 @@ See Cognee in action:
 
 ### Persistent Agent Memory
 
-[Cognee Memory for LangGraph Agents](https://github.com/user-attachments/assets/e113b628-7212-4a2b-b288-0be39a93a1c3)
-
-### Simple GraphRAG
-
-[Watch Demo](https://github.com/user-attachments/assets/f2186b2e-305a-42b0-9c2d-9f4473f15df8)
-
-### Cognee with Ollama
-
-[Watch Demo](https://github.com/user-attachments/assets/39672858-f774-4136-b957-1e2de67b8981)
+[![Watch Demo](https://img.youtube.com/vi/8hmqS2Y5RVQ/maxresdefault.jpg)](https://www.youtube.com/watch?v=8hmqS2Y5RVQ&t=13s)
 
 
 ## Community & Support

{cognee-0.5.4.dev1 → cognee-0.5.5.dev0}/cognee/api/client.py

@@ -38,6 +38,7 @@ from cognee.api.v1.users.routers import (
     get_users_router,
     get_visualize_router,
     get_configuration_router,
+    get_user_id_by_email_router,
 )
 from cognee.modules.users.methods.get_authenticated_user import REQUIRE_AUTHENTICATION
 
@@ -238,6 +239,12 @@ app.include_router(
     tags=["users"],
 )
 
+app.include_router(
+    get_user_id_by_email_router(),
+    prefix="/api/v1/users",
+    tags=["users"],
+)
+
 app.include_router(
     get_notebooks_router(),
     prefix="/api/v1/notebooks",

{cognee-0.5.4.dev1 → cognee-0.5.5.dev0}/cognee/api/v1/config/config.py

@@ -15,6 +15,24 @@ from cognee.api.v1.exceptions.exceptions import InvalidConfigAttributeError
 
 
 class config:
+    """
+    Configuration namespace for Cognee's runtime settings.
+
+    All methods are static and configure LLM providers, database backends,
+    chunking strategies, and storage paths at runtime without requiring
+    environment variable changes.
+
+    Example:
+        ```python
+        import cognee
+
+        cognee.config.set_llm_api_key("your-api-key")
+        cognee.config.set_llm_model("gpt-4o-mini")
+        cognee.config.system_root_directory("/path/to/system")
+        cognee.config.set_vector_db_provider("chromadb")
+        ```
+    """
+
     @staticmethod
     def system_root_directory(system_root_directory: str):
         base_config = get_base_config()

{cognee-0.5.4.dev1 → cognee-0.5.5.dev0}/cognee/api/v1/datasets/datasets.py

@@ -23,6 +23,27 @@ logger = get_logger()
 
 
 class datasets:
+    """
+    Dataset management namespace for Cognee.
+
+    All methods are static and provide operations for listing, inspecting,
+    and deleting datasets and the data items within them.
+
+    Example:
+        ```python
+        import cognee
+
+        # List all accessible datasets
+        all_datasets = await cognee.datasets.list_datasets()
+
+        # Check cognify processing status for datasets
+        status = await cognee.datasets.get_status([dataset_id])
+
+        # Delete a specific data item from a dataset
+        await cognee.datasets.delete_data(dataset_id=dataset_id, data_id=data_id)
+        ```
+    """
+
     @staticmethod
     async def list_datasets(user: Optional[User] = None):
         if user is None:

cognee-0.5.5.dev0/cognee/api/v1/delete/__init__.py

@@ -0,0 +1,39 @@
+from uuid import UUID
+from typing import Optional
+
+from deprecated import deprecated
+
+from cognee.api.v1.datasets import datasets
+from cognee.modules.users.models import User
+
+
+@deprecated(
+    reason="cognee.delete is deprecated. Use `datasets.delete_data` instead.", version="0.3.9"
+)
+async def delete(data_id: UUID, dataset_id: UUID, mode: str = "soft", user: Optional[User] = None):
+    """
+    Remove a specific data item from a dataset.
+
+    Deprecated since version 0.3.9. Use ``cognee.datasets.delete_data`` instead.
+
+    Args:
+        data_id: UUID of the data item to remove.
+        dataset_id: UUID of the dataset containing the data item.
+        mode: Deletion mode. Defaults to "soft".
+        user: User context for authentication and permissions. Uses default user if None.
+
+    Returns:
+        dict: Status of the deletion operation, e.g. ``{"status": "success"}``.
+
+    Example:
+        ```python
+        import cognee
+
+        # Preferred approach (not deprecated)
+        await cognee.datasets.delete_data(dataset_id=dataset_id, data_id=data_id)
+
+        # Legacy usage (deprecated since 0.3.9)
+        await cognee.delete(data_id=data_id, dataset_id=dataset_id)
+        ```
+    """
+    return await datasets.delete_data(data_id=data_id, dataset_id=dataset_id, mode=mode, user=user)

{cognee-0.5.4.dev1 → cognee-0.5.5.dev0}/cognee/api/v1/users/routers/__init__.py

@@ -5,3 +5,4 @@ from .get_users_router import get_users_router
 from .get_verify_router import get_verify_router
 from .get_visualize_router import get_visualize_router
 from .get_configuration_router import get_configuration_router
+from .get_user_id_by_email_router import get_user_id_by_email_router

cognee-0.5.5.dev0/cognee/api/v1/users/routers/get_user_id_by_email_router.py

@@ -0,0 +1,26 @@
+from fastapi import APIRouter, HTTPException, Depends
+from cognee.api.DTO import InDTO
+from pydantic import EmailStr
+
+from cognee.modules.users.models.User import User
+from cognee.modules.users.methods import get_authenticated_user
+from cognee.modules.users.methods.get_user_id_by_email import get_user_id_by_email
+
+
+class UserEmailRequest(InDTO):
+    email: EmailStr
+
+
+def get_user_id_by_email_router() -> APIRouter:
+    router = APIRouter()
+
+    @router.post("/get-user-id")
+    async def get_user_id(body: UserEmailRequest, user: User = Depends(get_authenticated_user)):
+        user_id = await get_user_id_by_email(str(body.email))
+
+        if user_id is None:
+            raise HTTPException(status_code=404, detail="User not found")
+
+        return {"user_id": str(user_id)}
+
+    return router

{cognee-0.5.4.dev1 → cognee-0.5.5.dev0}/cognee/context_global_variables.py

@@ -138,8 +138,6 @@ async def set_database_global_context_variables(dataset: Union[str, UUID], user_
     )
 
     # Set vector and graph database configuration based on dataset database information
-    # TODO: Add better handling of vector and graph config across Cognee.
-    #  LRU_CACHE takes into account order of inputs, if order of inputs is changed it will be registered as a new DB adapter
     vector_config = {
         "vector_db_provider": dataset_database.vector_database_provider,
         "vector_db_url": dataset_database.vector_database_url,

{cognee-0.5.4.dev1 → cognee-0.5.5.dev0}/cognee/infrastructure/databases/cache/cache_db_interface.py

@@ -75,11 +75,13 @@ class CacheDBInterface(ABC):
         feedback_text: str | None = None,
         feedback_score: int | None = None,
         used_graph_element_ids: dict | None = None,
+        memify_metadata: dict | None = None,
     ):
         """
         Add a Q/A/context triplet to a cache session.
         Uses the same QA fields as update_qa_entry for consistent structure.
         used_graph_element_ids: Optional dict with keys "node_ids" and "edge_ids" (lists of str).
+        memify_metadata: Optional dict with status keys (e.g. "feedback_weights_applied") and bool values.
         """
         pass
 
@@ -117,6 +119,7 @@ class CacheDBInterface(ABC):
         feedback_text: str | None = None,
         feedback_score: int | None = None,
         used_graph_element_ids: dict | None = None,
+        memify_metadata: dict | None = None,
     ) -> bool:
         """
         Update a QA entry by qa_id. Same QA fields as create_qa_entry.

{cognee-0.5.4.dev1 → cognee-0.5.5.dev0}/cognee/infrastructure/databases/cache/fscache/FsCacheAdapter.py

@@ -45,6 +45,7 @@ class FSCacheAdapter(CacheDBInterface):
         feedback_text: str | None = None,
         feedback_score: int | None = None,
         used_graph_element_ids: dict | None = None,
+        memify_metadata: dict | None = None,
     ) -> dict:
         entry = SessionQAEntry(
             time=datetime.utcnow().isoformat(),
@@ -55,6 +56,7 @@ class FSCacheAdapter(CacheDBInterface):
             feedback_text=feedback_text,
             feedback_score=feedback_score,
             used_graph_element_ids=used_graph_element_ids,
+            memify_metadata=memify_metadata,
         )
         return entry.model_dump()
 
@@ -79,6 +81,7 @@ class FSCacheAdapter(CacheDBInterface):
         feedback_text: str | None = None,
         feedback_score: int | None = None,
         used_graph_element_ids: dict | None = None,
+        memify_metadata: dict | None = None,
     ) -> dict:
         merged = {**entry}
         if question is not None:
@@ -93,6 +96,13 @@ class FSCacheAdapter(CacheDBInterface):
             merged["feedback_score"] = feedback_score
         if used_graph_element_ids is not None:
             merged["used_graph_element_ids"] = used_graph_element_ids
+        if memify_metadata is not None:
+            existing_metadata = merged.get("memify_metadata")
+            if isinstance(existing_metadata, dict):
+                merged["memify_metadata"] = {**existing_metadata, **memify_metadata}
+            else:
+                merged["memify_metadata"] = memify_metadata
+
         return merged
 
     @staticmethod
@@ -138,6 +148,7 @@ class FSCacheAdapter(CacheDBInterface):
         feedback_text: str | None = None,
         feedback_score: int | None = None,
         used_graph_element_ids: dict | None = None,
+        memify_metadata: dict | None = None,
     ):
         try:
             session_key = self._session_key(user_id, session_id)
@@ -149,6 +160,7 @@ class FSCacheAdapter(CacheDBInterface):
                 feedback_text,
                 feedback_score,
                 used_graph_element_ids=used_graph_element_ids,
+                memify_metadata=memify_metadata,
             )
             entries = self._load_entries(session_key)
             entries.append(qa_entry)
@@ -180,6 +192,7 @@ class FSCacheAdapter(CacheDBInterface):
         feedback_text: str | None = None,
         feedback_score: int | None = None,
         used_graph_element_ids: dict | None = None,
+        memify_metadata: dict | None = None,
     ) -> bool:
         """
         Update a QA entry by qa_id. Same QA fields as create_qa_entry.
@@ -200,6 +213,7 @@ class FSCacheAdapter(CacheDBInterface):
                 feedback_text,
                 feedback_score,
                 used_graph_element_ids=used_graph_element_ids,
+                memify_metadata=memify_metadata,
             )
             entries[idx] = self._validate_entry_dict(merged)
             self._save_entries(session_key, entries)

{cognee-0.5.4.dev1 → cognee-0.5.5.dev0}/cognee/infrastructure/databases/cache/models.py

@@ -24,6 +24,7 @@ class SessionQAEntry(BaseModel):
         feedback_text: Optional user feedback text.
         feedback_score: Optional feedback score 1-5.
         used_graph_element_ids: Optional dict with only "node_ids" and "edge_ids" (lists of str).
+        memify_metadata: Optional dict with memify status keys (e.g. "feedback_weights_applied") and bool values.
     """
 
     time: str
@@ -34,6 +35,7 @@ class SessionQAEntry(BaseModel):
     feedback_text: Optional[str] = None
     feedback_score: Optional[int] = None
     used_graph_element_ids: Optional[Dict[str, List[str]]] = None
+    memify_metadata: Optional[Dict[str, bool]] = None
 
     @field_validator("used_graph_element_ids")
     @classmethod
@@ -60,3 +62,19 @@ class SessionQAEntry(BaseModel):
         if v is not None and (v < 1 or v > 5):
             raise ValueError("feedback_score must be between 1 and 5")
         return v
+
+    @field_validator("memify_metadata")
+    @classmethod
+    def memify_metadata_only_pipeline_keys(
+        cls, v: Optional[Dict[str, bool]]
+    ) -> Optional[Dict[str, bool]]:
+        if v is None:
+            return None
+        if not isinstance(v, dict):
+            raise ValueError("memify_metadata must be a dict or None")
+        out: Dict[str, bool] = {}
+        for key, val in v.items():
+            if not isinstance(key, str) or not isinstance(val, bool):
+                raise ValueError("memify_metadata may only have string keys and bool values")
+            out[key] = val
+        return out if out else None

{cognee-0.5.4.dev1 → cognee-0.5.5.dev0}/cognee/infrastructure/databases/cache/redis/RedisAdapter.py

@@ -91,6 +91,7 @@ class RedisAdapter(CacheDBInterface):
         feedback_text: str | None = None,
         feedback_score: int | None = None,
         used_graph_element_ids: dict | None = None,
+        memify_metadata: dict | None = None,
     ) -> dict:
         entry = SessionQAEntry(
             time=datetime.utcnow().isoformat(),
@@ -101,6 +102,7 @@ class RedisAdapter(CacheDBInterface):
             feedback_text=feedback_text,
             feedback_score=feedback_score,
             used_graph_element_ids=used_graph_element_ids,
+            memify_metadata=memify_metadata,
         )
         return entry.model_dump()
 
@@ -125,6 +127,7 @@ class RedisAdapter(CacheDBInterface):
         feedback_text: str | None = None,
         feedback_score: int | None = None,
         used_graph_element_ids: dict | None = None,
+        memify_metadata: dict | None = None,
     ) -> dict:
         merged = {**entry}
         if question is not None:
@@ -139,6 +142,12 @@ class RedisAdapter(CacheDBInterface):
             merged["feedback_score"] = feedback_score
         if used_graph_element_ids is not None:
             merged["used_graph_element_ids"] = used_graph_element_ids
+        if memify_metadata is not None:
+            existing_metadata = merged.get("memify_metadata")
+            if isinstance(existing_metadata, dict):
+                merged["memify_metadata"] = {**existing_metadata, **memify_metadata}
+            else:
+                merged["memify_metadata"] = memify_metadata
         return merged
 
     @staticmethod
@@ -210,6 +219,7 @@ class RedisAdapter(CacheDBInterface):
         feedback_text: str | None = None,
         feedback_score: int | None = None,
         used_graph_element_ids: dict | None = None,
+        memify_metadata: dict | None = None,
     ):
         """
         Add a Q/A/context triplet to a Redis list for this session.
@@ -225,6 +235,7 @@ class RedisAdapter(CacheDBInterface):
                 feedback_text,
                 feedback_score,
                 used_graph_element_ids=used_graph_element_ids,
+                memify_metadata=memify_metadata,
             )
             await self.async_redis.rpush(session_key, json.dumps(qa_entry))
         except (redis.ConnectionError, redis.TimeoutError) as e:
@@ -265,6 +276,7 @@ class RedisAdapter(CacheDBInterface):
         feedback_text: str | None = None,
         feedback_score: int | None = None,
         used_graph_element_ids: dict | None = None,
+        memify_metadata: dict | None = None,
     ) -> bool:
         """
         Update a QA entry by qa_id. Same QA fields as create_qa_entry.
@@ -285,6 +297,7 @@ class RedisAdapter(CacheDBInterface):
                 feedback_text,
                 feedback_score,
                 used_graph_element_ids=used_graph_element_ids,
+                memify_metadata=memify_metadata,
             )
             entries[idx] = self._validate_entry_dict(merged)
             await self._write_entry_at(session_key, idx, entries[idx])

{cognee-0.5.4.dev1 → cognee-0.5.5.dev0}/cognee/infrastructure/databases/graph/graph_db_interface.py

@@ -291,3 +291,35 @@ class GraphDBInterface(ABC):
               are lists of attribute values to filter by.
         """
         raise NotImplementedError
+
+    async def get_node_feedback_weights(self, node_ids: List[str]) -> Dict[str, float]:
+        """
+        Retrieve node feedback weights for multiple node ids.
+        Returns only found node ids.
+        """
+        raise NotImplementedError("get_node_feedback_weights is not implemented for this adapter")
+
+    async def set_node_feedback_weights(
+        self, node_feedback_weights: Dict[str, float]
+    ) -> Dict[str, bool]:
+        """
+        Persist node feedback weights for multiple node ids.
+        Returns per-id update success.
+        """
+        raise NotImplementedError("set_node_feedback_weights is not implemented for this adapter")
+
+    async def get_edge_feedback_weights(self, edge_object_ids: List[str]) -> Dict[str, float]:
+        """
+        Retrieve edge feedback weights for multiple edge_object_ids.
+        Returns only found edge ids.
+        """
+        raise NotImplementedError("get_edge_feedback_weights is not implemented for this adapter")
+
+    async def set_edge_feedback_weights(
+        self, edge_feedback_weights: Dict[str, float]
+    ) -> Dict[str, bool]:
+        """
+        Persist edge feedback weights for multiple edge_object_ids.
+        Returns per-id update success.
+        """
+        raise NotImplementedError("set_edge_feedback_weights is not implemented for this adapter")