cognee 1.2.0.dev0__tar.gz → 1.2.0.dev1__tar.gz

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cognee
-Version: 1.2.0.dev0
+Version: 1.2.0.dev1
 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
@@ -30,7 +30,7 @@ Requires-Dist: fastapi<1.0.0,>=0.116.2
 Requires-Dist: filelock<4.0.0,>=3.12.0
 Requires-Dist: filetype<2.0.0,>=1.2.0
 Requires-Dist: gunicorn<24,>=20.1.0
-Requires-Dist: instructor<1.15.3,>=1.9.1
+Requires-Dist: instructor<2.0.0,>=1.9.1
 Requires-Dist: jinja2<4,>=3.1.3
 Requires-Dist: ladybug<0.18,>=0.16.0
 Requires-Dist: lancedb<1.0.0,>=0.24.3
@@ -185,7 +185,7 @@ Description-Content-Type: text/markdown
 
   <br />
 
-  Cognee - The Brain behind your Agents
+  Cognee - The Open-Source AI Memory Platform for Agents
 
   <p align="center">
   <a href="https://www.youtube.com/watch?v=8hmqS2Y5RVQ&t=13s">Demo</a>
@@ -217,10 +217,10 @@ Description-Content-Type: text/markdown
   </a>
 </p>
 
-Cognee gives AI agents a shared, improving memory of your data, decisions, and workflows so they can recall, connect, and act with context.
+Cognee is the open-source AI memory platform that gives AI agents persistent long-term memory across sessions. Ingest data in any format, build a self-hosted knowledge graph, and let every agent recall, connect, and act with full context
 
   <p align="center">
-  🌐 Available Languages
+  🌐 This README is also available in:
   :
   <!-- Keep these links. Translations will automatically update with the README. -->
   <a href="https://www.readme-i18n.com/topoteretes/cognee?lang=de">Deutsch</a> |
@@ -239,14 +239,12 @@ Cognee gives AI agents a shared, improving memory of your data, decisions, and w
 </div>
 </div>
 
-
+📄 Read the research paper: [Optimizing the Interface Between Knowledge Graphs and LLMs for Complex Reasoning](https://arxiv.org/abs/2505.24478) — Markovic et al., 2025
 
 
 ## About Cognee
 
-Cognee is an open-source memory control plane for your Agents that lets you ingest data in any format or structure and continuously learns to provide the right context. It combines embeddings, graphs and cognitive science approaches to make your documents both searchable by meaning and connected by relationships as they change and evolve.
-
-
+Cognee is an open-source AI memory platform for AI Agents. Ingest data in any format, and Cognee continuously builds a self-hosted knowledge graph that gives your agents persistent long-term memory across sessions. Cognee combines vector embeddings, graph reasoning, and cognitive-science-grounded ontology generation to make documents both searchable by meaning and connected by relationships that evolve as your knowledge does.
 
 :star: _Help us reach more developers and grow the cognee community. Star this repo!_
 

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/README.md

@@ -5,7 +5,7 @@
 
   <br />
 
-  Cognee - The Brain behind your Agents
+  Cognee - The Open-Source AI Memory Platform for Agents
 
   <p align="center">
   <a href="https://www.youtube.com/watch?v=8hmqS2Y5RVQ&t=13s">Demo</a>
@@ -37,10 +37,10 @@
   </a>
 </p>
 
-Cognee gives AI agents a shared, improving memory of your data, decisions, and workflows so they can recall, connect, and act with context.
+Cognee is the open-source AI memory platform that gives AI agents persistent long-term memory across sessions. Ingest data in any format, build a self-hosted knowledge graph, and let every agent recall, connect, and act with full context
 
   <p align="center">
-  🌐 Available Languages
+  🌐 This README is also available in:
   :
   <!-- Keep these links. Translations will automatically update with the README. -->
   <a href="https://www.readme-i18n.com/topoteretes/cognee?lang=de">Deutsch</a> |
@@ -59,14 +59,12 @@ Cognee gives AI agents a shared, improving memory of your data, decisions, and w
 </div>
 </div>
 
-
+📄 Read the research paper: [Optimizing the Interface Between Knowledge Graphs and LLMs for Complex Reasoning](https://arxiv.org/abs/2505.24478) — Markovic et al., 2025
 
 
 ## About Cognee
 
-Cognee is an open-source memory control plane for your Agents that lets you ingest data in any format or structure and continuously learns to provide the right context. It combines embeddings, graphs and cognitive science approaches to make your documents both searchable by meaning and connected by relationships as they change and evolve.
-
-
+Cognee is an open-source AI memory platform for AI Agents. Ingest data in any format, and Cognee continuously builds a self-hosted knowledge graph that gives your agents persistent long-term memory across sessions. Cognee combines vector embeddings, graph reasoning, and cognitive-science-grounded ontology generation to make documents both searchable by meaning and connected by relationships that evolve as your knowledge does.
 
 :star: _Help us reach more developers and grow the cognee community. Star this repo!_
 

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/cognee/api/v1/add/add.py

@@ -1,6 +1,3 @@
-import inspect
-from tempfile import SpooledTemporaryFile
-from types import SimpleNamespace
 from uuid import UUID
 from typing import Union, BinaryIO, List, Optional, Any
 
@@ -19,78 +16,12 @@ from cognee.modules.engine.operations.setup import setup
 from cognee.tasks.ingestion import ingest_data, resolve_data_directories
 from cognee.tasks.ingestion.data_item import DataItem
 from cognee.tasks.ingestion.resolve_dlt_sources import resolve_dlt_sources
+from cognee.tasks.ingestion.utils import materialize_stream_for_background
 from cognee.shared.logging_utils import get_logger
 
 logger = get_logger()
 
 
-def _normalize_filename(filename: Optional[str], index: int) -> str:
-    if not filename:
-        return f"upload_{index}.bin"
-    normalized = str(filename).replace("\\", "/").split("/")[-1]
-    return normalized or f"upload_{index}.bin"
-
-
-async def _read_stream_bytes(stream: Any) -> bytes:
-    if not hasattr(stream, "read"):
-        raise TypeError(f"Expected stream-like object, got: {type(stream)}")
-
-    # Best effort to read from the start of the stream.
-    if hasattr(stream, "seek"):
-        try:
-            stream.seek(0)
-        except Exception:
-            pass
-
-    data = stream.read()
-    if inspect.isawaitable(data):
-        data = await data
-
-    if isinstance(data, str):
-        data = data.encode("utf-8")
-    if not isinstance(data, (bytes, bytearray)):
-        raise TypeError(f"Unsupported stream payload type: {type(data)}")
-
-    return bytes(data)
-
-
-async def _materialize_stream_for_background(data_item: Any, index: int = 0) -> Any:
-    if isinstance(data_item, DataItem):
-        return DataItem(
-            data=await _materialize_stream_for_background(data_item.data, index=index),
-            label=data_item.label,
-            external_metadata=data_item.external_metadata,
-            data_id=data_item.data_id,
-        )
-
-    if isinstance(data_item, list):
-        return [
-            await _materialize_stream_for_background(item, index=i)
-            for i, item in enumerate(data_item)
-        ]
-
-    # Keep stable primitives untouched.
-    if isinstance(data_item, str):
-        return data_item
-
-    stream = getattr(data_item, "file", data_item if hasattr(data_item, "read") else None)
-    if stream is None:
-        return data_item
-
-    payload = await _read_stream_bytes(stream)
-    buffer = SpooledTemporaryFile(mode="w+b")
-    buffer.write(payload)
-    buffer.seek(0)
-
-    filename = _normalize_filename(
-        getattr(data_item, "filename", None) or getattr(stream, "name", None),
-        index=index,
-    )
-
-    # Ingestion path supports objects exposing `.file` and `.filename`.
-    return SimpleNamespace(file=buffer, filename=filename)
-
-
 async def add(
     data: Union[
         BinaryIO,
@@ -315,7 +246,7 @@ async def add(
         if orphan_cleanup is not None:
             await orphan_cleanup()
             orphan_cleanup = None
-        data = await _materialize_stream_for_background(data)
+        data = await materialize_stream_for_background(data)
 
     await reset_dataset_pipeline_run_status(
         authorized_dataset.id, user, pipeline_names=["add_pipeline", "cognify_pipeline"]

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/cognee/api/v1/remember/remember.py

@@ -195,11 +195,17 @@ async def _remember_entry(
         result = RememberResult(
             status=payload.get("status", "session_stored"),
             dataset_name=payload.get("dataset_name", dataset_name),
+            dataset_id=payload.get("dataset_id"),
             session_ids=payload.get("session_ids"),
+            pipeline_run_id=payload.get("pipeline_run_id"),
         )
         result.entry_type = payload.get("entry_type")
         result.entry_id = payload.get("entry_id")
         result.elapsed_seconds = payload.get("elapsed_seconds")
+        result.items_processed = payload.get("items_processed", 0)
+        result.items = payload.get("items", []) or []
+        result.content_hash = payload.get("content_hash")
+        result.raw_result = payload
         if payload.get("error"):
             result.error = payload["error"]
         return result
@@ -853,6 +859,7 @@ async def _remember_inner(
         )
 
     if content_type == "skills":
+        import shutil
         import tempfile
         from pathlib import Path as _Path
 
@@ -907,7 +914,20 @@ async def _remember_inner(
                 if isinstance(payload, str):
                     payload = payload.encode("utf-8")
                 dest.write_bytes(payload or b"")
-            skill_source = tmp_root
+            skill_files = list(tmp_root.rglob("SKILL.md"))
+            if skill_files:
+                skill_source = tmp_root
+            else:
+                raw_files = [path for path in tmp_root.rglob("*") if path.is_file()]
+                if raw_files:
+                    canonical_root = tmp_root / "__canonical_skills__"
+                    for index, raw_file in enumerate(raw_files):
+                        target_dir = canonical_root / f"skill-{index:04d}"
+                        target_dir.mkdir(parents=True, exist_ok=True)
+                        shutil.copy2(raw_file, target_dir / "SKILL.md")
+                    skill_source = canonical_root
+                else:
+                    skill_source = tmp_root
 
         try:
             async with set_database_global_context_variables(dataset.id, owner_id):
@@ -1063,6 +1083,11 @@ async def _remember_inner(
             await improve(**improve_kwargs)
 
     if run_in_background:
+        # Background runs must not depend on caller/request-scoped stream lifetimes.
+        # Materialize stream-like inputs into owned in-memory buffers up front.
+        from cognee.tasks.ingestion.utils import materialize_stream_for_background
+
+        data = await materialize_stream_for_background(data)
 
         async def _remember_background():
             try:

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/cognee/infrastructure/databases/graph/postgres/adapter.py

@@ -527,6 +527,52 @@ class PostgresAdapter(GraphDBInterface):
 
             return nodes, edges
 
+    async def get_id_filtered_graph_data(
+        self, target_ids: List[str]
+    ) -> Tuple[List[Tuple[str, Dict[str, Any]]], List[Tuple[str, str, str, Dict[str, Any]]]]:
+        """Retrieve the subgraph touching target_ids: edges with either endpoint
+        in the set, plus all endpoint nodes of those edges (edge-driven,
+        matching the Ladybug/Neo4j contract). Lets CogneeGraph project only the
+        vector-search neighborhood instead of the full graph.
+        """
+        if not target_ids:
+            return [], []
+        ids = [str(i) for i in target_ids]
+
+        async with self._session() as session:
+            edge_result = await session.execute(
+                text("""
+                    SELECT source_id, target_id, relationship_name, properties
+                    FROM graph_edge
+                    WHERE source_id = ANY(:ids) OR target_id = ANY(:ids)
+                """),
+                {"ids": ids},
+            )
+            edges = []
+            endpoint_ids = set()
+            for row in edge_result.fetchall():
+                props = {}
+                if row[3]:
+                    props = row[3] if isinstance(row[3], dict) else json.loads(row[3])
+                endpoint_ids.update((row[0], row[1]))
+                edges.append((row[0], row[1], row[2], props))
+
+            if not endpoint_ids:
+                return [], []
+
+            node_result = await session.execute(
+                text("SELECT id, name, type, properties FROM graph_node WHERE id = ANY(:ids)"),
+                {"ids": list(endpoint_ids)},
+            )
+            nodes = []
+            for row in node_result.fetchall():
+                data = {"name": row[1], "type": row[2]}
+                if row[3]:
+                    data.update(row[3] if isinstance(row[3], dict) else json.loads(row[3]))
+                nodes.append((row[0], data))
+
+            return nodes, edges
+
     async def get_filtered_graph_data(
         self, attribute_filters: List[Dict[str, List[Union[str, int]]]]
     ) -> Tuple[List[Tuple[str, Dict]], List[Tuple[str, str, str, Dict]]]:

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/cognee/infrastructure/databases/vector/lancedb/LanceDBAdapter.py

@@ -906,6 +906,10 @@ class LanceDBAdapter(VectorDBInterface):
 
     async def retrieve(self, collection_name: str, data_point_ids: list[str]):
         """Return rows from `collection_name` whose id is in `data_point_ids`."""
+        if not data_point_ids:
+            # No ids requested. Avoid building an "id IN ()" filter, which lance
+            # rejects as a parse error; pgvector/chromadb return [] here too.
+            return []
         try:
             collection = await self.get_collection(collection_name)
         except CollectionNotFoundError:

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/cognee/infrastructure/files/utils/get_file_metadata.py

@@ -47,6 +47,7 @@ async def get_file_metadata(file: BinaryIO, name: str | None = None) -> FileMeta
         - FileMetadata: A dictionary containing the file's name, path, MIME type, file
           extension, and content hash.
     """
+    content_hash = ""
     try:
         file.seek(0)
         content_hash = await get_file_content_hash(file)

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/cognee/infrastructure/files/utils/guess_file_type.py

@@ -57,8 +57,19 @@ def guess_file_type(file: BinaryIO, name: str | None = None) -> filetype.Type:
         return file_type
 
     if ext in [".csv"]:
-        file_type = Type("text/csv", "csv")
-        return file_type
+        return Type("text/csv", "csv")
+
+    if ext in [".md", ".markdown"]:
+        return Type("text/markdown", "md")
+
+    if ext in [".json"]:
+        return Type("application/json", "json")
+
+    if ext in [".xml"]:
+        return Type("application/xml", "xml")
+
+    if ext in [".yaml", ".yml"]:
+        return Type("application/yaml", "yaml")
 
     file_type = filetype.guess(file)
 
@@ -66,7 +77,4 @@ def guess_file_type(file: BinaryIO, name: str | None = None) -> filetype.Type:
     if file_type is None:
         file_type = Type("text/plain", "txt")
 
-    if file_type is None:
-        raise FileTypeException(f"Unknown file detected: {file.name}.")
-
     return file_type

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/cognee/modules/pipelines/operations/run_tasks.py

@@ -129,7 +129,6 @@ async def run_tasks(
 
             gathered = await asyncio.gather(
                 *[asyncio.create_task(_run_item(item)) for item in data],
-                return_exceptions=True,
             )
 
             # Separate successes from unhandled exceptions

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/cognee/modules/users/methods/get_authenticated_user.py

@@ -4,6 +4,7 @@ from fastapi import Depends, HTTPException
 from ..models import User
 from ..get_fastapi_users import get_fastapi_users
 from .get_default_user import get_default_user
+from .get_user import get_user
 from cognee.shared.logging_utils import get_logger
 
 
@@ -99,5 +100,11 @@ async def get_authenticated_user(
             raise HTTPException(
                 status_code=500, detail=f"Failed to create default user: {str(e)}"
             ) from e
+    else:
+        # FastAPI Users returns a session-bound instance without eager-loaded
+        # relationships. Re-fetch so callers (including background tasks that
+        # outlive the request) can access user.tenants / user.roles without
+        # DetachedInstanceError.
+        user = await get_user(user.id)
 
     return user

cognee-1.2.0.dev1/cognee/tasks/ingestion/utils.py

@@ -0,0 +1,73 @@
+import inspect
+from tempfile import SpooledTemporaryFile
+from types import SimpleNamespace
+from typing import Any, Optional
+
+from cognee.tasks.ingestion.data_item import DataItem
+
+
+def _normalize_filename(filename: Optional[str], index: int) -> str:
+    if not filename:
+        return f"upload_{index}.bin"
+    normalized = str(filename).replace("\\", "/").split("/")[-1]
+    return normalized or f"upload_{index}.bin"
+
+
+async def _read_stream_bytes(stream: Any) -> bytes:
+    if not hasattr(stream, "read"):
+        raise TypeError(f"Expected stream-like object, got: {type(stream)}")
+
+    # Best effort to read from the start of the stream.
+    if hasattr(stream, "seek"):
+        try:
+            stream.seek(0)
+        except Exception:
+            pass
+
+    data = stream.read()
+    if inspect.isawaitable(data):
+        data = await data
+
+    if isinstance(data, str):
+        data = data.encode("utf-8")
+    if not isinstance(data, (bytes, bytearray)):
+        raise TypeError(f"Unsupported stream payload type: {type(data)}")
+
+    return bytes(data)
+
+
+async def materialize_stream_for_background(data_item: Any, index: int = 0) -> Any:
+    if isinstance(data_item, DataItem):
+        return DataItem(
+            data=await materialize_stream_for_background(data_item.data, index=index),
+            label=data_item.label,
+            external_metadata=data_item.external_metadata,
+            data_id=data_item.data_id,
+        )
+
+    if isinstance(data_item, list):
+        return [
+            await materialize_stream_for_background(item, index=i)
+            for i, item in enumerate(data_item)
+        ]
+
+    # Keep stable primitives untouched.
+    if isinstance(data_item, str):
+        return data_item
+
+    stream = getattr(data_item, "file", data_item if hasattr(data_item, "read") else None)
+    if stream is None:
+        return data_item
+
+    payload = await _read_stream_bytes(stream)
+    buffer = SpooledTemporaryFile(mode="w+b")
+    buffer.write(payload)
+    buffer.seek(0)
+
+    filename = _normalize_filename(
+        getattr(data_item, "filename", None) or getattr(stream, "name", None),
+        index=index,
+    )
+
+    # Ingestion path supports objects exposing `.file` and `.filename`.
+    return SimpleNamespace(file=buffer, filename=filename)

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/cognee/tests/performance/statistics_percentile/bench_cognee.py

@@ -58,8 +58,14 @@ def _resolve_config(args: argparse.Namespace) -> dict:
     if not api_key and not mock_llm:
         sys.exit("Error: LLM_API_KEY is not set (CLI, .env, or environment)")
 
+    # Embeddings may use a different provider/key than the LLM (e.g.
+    # + OpenAI embeddings). Resolve the embedding key independently, falling back
+    # to the LLM/OpenAI key when LLM and embeddings share a provider.
+    embedding_api_key = pick(None, "EMBEDDING_API_KEY", "") or api_key
+
     return {
         "api_key": api_key or "mock-key",
+        "embedding_api_key": embedding_api_key or "mock-key",
         "llm_provider": pick(args.llm_provider, "LLM_PROVIDER", DEFAULT_LLM_PROVIDER),
         "llm_model": pick(args.llm_model, "LLM_MODEL", DEFAULT_LLM_MODEL),
         "embedding_provider": pick(
@@ -169,6 +175,18 @@ async def run_benchmark(
 ) -> dict:
     import cognee
 
+    # Register community adapters before any engine is created. Comma-separated
+    # module names; a module-level register() is called if present (some
+    # adapters register on import alone).
+    import importlib
+
+    for module_name in filter(None, os.environ.get("COGNEE_REGISTER_ADAPTERS", "").split(",")):
+        module = importlib.import_module(module_name)
+        register = getattr(module, "register", None)
+        if callable(register):
+            register()
+        print(f"Registered adapter module: {module_name}")
+
     llm_model = config["llm_model"]
     llm_provider = config["llm_provider"]
     embedding_model = config["embedding_model"]
@@ -180,7 +198,7 @@ async def run_benchmark(
     cognee.config.set_embedding_provider(config["embedding_provider"])
     cognee.config.set_embedding_model(embedding_model)
     cognee.config.set_embedding_dimensions(embedding_dims)
-    cognee.config.set_embedding_api_key(config["api_key"])
+    cognee.config.set_embedding_api_key(config["embedding_api_key"])
 
     if config.get("mock_llm"):
         mock_data = _load_mock_data(config["mock_memories_file"])

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/cognee/tests/performance/statistics_percentile_report.py

@@ -14,6 +14,8 @@ from __future__ import annotations
 
 import argparse
 import json
+import os
+import shlex
 import subprocess
 import sys
 import tempfile
@@ -60,7 +62,11 @@ def run_single(run_num: int, total: int, extra_args: list[str]) -> dict:
     with tempfile.NamedTemporaryFile(suffix=".json", delete=False) as tmp:
         tmp_path = tmp.name
 
-    cmd = [sys.executable, str(BENCH_SCRIPT), "--output", tmp_path] + extra_args
+    # BENCH_CMD lets another SDK (e.g. the Rust cognee-cli bench) drive this
+    # orchestrator unchanged. When unset, fall back to the Python bench script.
+    bench_cmd = os.environ.get("BENCH_CMD")
+    base = shlex.split(bench_cmd) if bench_cmd else [sys.executable, str(BENCH_SCRIPT)]
+    cmd = base + ["--output", tmp_path] + extra_args
     print(f"\n{'=' * 60}")
     print(f"  RUN {run_num}/{total}")
     print(f"{'=' * 60}\n")

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/cognee/tests/unit/infrastructure/databases/vector/test_lancedb_lifecycle.py

@@ -112,6 +112,39 @@ async def test_close_is_idempotent(tmp_path):
     assert adapter._permanently_closed is True
 
 
+@pytest.mark.asyncio
+@pytest.mark.skipif(not HAS_LANCEDB, reason="lancedb not installed")
+async def test_retrieve_empty_ids_returns_empty(monkeypatch, tmp_path):
+    """retrieve() with no ids must return [] rather than build an ``id IN ()``
+    filter, which lance rejects as a parse error (parity with pgvector/chromadb;
+    e.g. has_new_chunks() passes an empty id list when data_chunks is empty)."""
+
+    class _FakeQuery:
+        def where(self, *_args, **_kwargs):
+            return self
+
+        async def to_list(self):
+            # Rows the filter path would surface; the empty-id guard must
+            # short-circuit before we ever reach here.
+            return [{"id": "00000000-0000-0000-0000-000000000000", "payload": {"text": "x"}}]
+
+    class _FakeCollection:
+        def query(self):
+            return _FakeQuery()
+
+    async def _fake_get_collection(_name):
+        return _FakeCollection()
+
+    adapter = LanceDBAdapter(
+        url=str(tmp_path / "lance_db"),
+        api_key=None,
+        embedding_engine=_FakeEmbeddingEngine(),
+    )
+    monkeypatch.setattr(adapter, "get_collection", _fake_get_collection)
+
+    assert await adapter.retrieve("any_collection", []) == []
+
+
 @pytest.mark.asyncio
 @pytest.mark.skipif(not HAS_LANCEDB, reason="lancedb not installed")
 async def test_concurrent_first_get_connection_returns_same_object(monkeypatch, tmp_path):

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/cognee/tests/unit/modules/users/test_conditional_authentication.py

@@ -35,9 +35,10 @@ class TestConditionalAuthentication:
         mock_get_default.assert_called_once()
 
     @pytest.mark.asyncio
+    @patch.object(gau_mod, "get_user", new_callable=AsyncMock)
     @patch.object(gau_mod, "get_default_user", new_callable=AsyncMock)
     async def test_require_authentication_false_with_valid_user_returns_user(
-        self, mock_get_default
+        self, mock_get_default, mock_get_user
     ):
         """Test that when REQUIRE_AUTHENTICATION=false and valid user, returns that user."""
         mock_authenticated_user = User(
@@ -47,8 +48,7 @@ class TestConditionalAuthentication:
             is_active=True,
             is_verified=True,
         )
-
-        # Use gau_mod.get_authenticated_user instead
+        mock_get_user.return_value = mock_authenticated_user
 
         # Test with authenticated user
         result = await gau_mod.get_authenticated_user(user=mock_authenticated_user)
@@ -57,8 +57,11 @@ class TestConditionalAuthentication:
         mock_get_default.assert_not_called()
 
     @pytest.mark.asyncio
+    @patch.object(gau_mod, "get_user", new_callable=AsyncMock)
     @patch.object(gau_mod, "get_default_user", new_callable=AsyncMock)
-    async def test_require_authentication_true_with_user_returns_user(self, mock_get_default):
+    async def test_require_authentication_true_with_user_returns_user(
+        self, mock_get_default, mock_get_user
+    ):
         """Test that when REQUIRE_AUTHENTICATION=true and user present, returns user."""
         mock_authenticated_user = User(
             id=uuid4(),
@@ -67,8 +70,7 @@ class TestConditionalAuthentication:
             is_active=True,
             is_verified=True,
         )
-
-        # Use gau_mod.get_authenticated_user instead
+        mock_get_user.return_value = mock_authenticated_user
 
         result = await gau_mod.get_authenticated_user(user=mock_authenticated_user)
 
@@ -213,8 +215,9 @@ class TestConditionalAuthenticationEdgeCases:
             await gau_mod.get_authenticated_user(user=None)
 
     @pytest.mark.asyncio
+    @patch.object(gau_mod, "get_user", new_callable=AsyncMock)
     @patch.object(gau_mod, "get_default_user", new_callable=AsyncMock)
-    async def test_user_type_consistency(self, mock_get_default):
+    async def test_user_type_consistency(self, mock_get_default, mock_get_user):
         """Test that the function always returns the same type."""
         mock_user = User(
             id=uuid4(),
@@ -223,6 +226,7 @@ class TestConditionalAuthenticationEdgeCases:
             is_active=True,
             is_verified=True,
         )
+        mock_get_user.return_value = mock_user
 
         mock_default_user = SimpleNamespace(id=uuid4(), email="default@example.com", is_active=True)
         mock_get_default.return_value = mock_default_user
@@ -270,7 +274,8 @@ class TestAuthenticationScenarios:
         assert result == mock_default_user
         mock_get_default.assert_called_once()
 
-    async def test_scenario_valid_active_user(self):
+    @patch.object(gau_mod, "get_user", new_callable=AsyncMock)
+    async def test_scenario_valid_active_user(self, mock_get_user):
         """Scenario: Valid JWT and user exists and is active → returns the user."""
         mock_user = User(
             id=uuid4(),
@@ -279,8 +284,7 @@ class TestAuthenticationScenarios:
             is_active=True,
             is_verified=True,
         )
-
-        # Use gau_mod.get_authenticated_user instead
+        mock_get_user.return_value = mock_user
 
         result = await gau_mod.get_authenticated_user(user=mock_user)
         assert result == mock_user

{cognee-1.2.0.dev0 → cognee-1.2.0.dev1}/pyproject.toml

@@ -1,7 +1,7 @@
 [project]
 name = "cognee"
 
-version = "1.2.0.dev0"
+version = "1.2.0.dev1"
 description = "Cognee - is a library for enriching LLM context with a semantic layer for better understanding and reasoning."
 authors = [
     { name = "Vasilije Markovic" },
@@ -31,7 +31,7 @@ dependencies = [
     "filelock>=3.12.0,<4.0.0",  # cross-process migration lock for SQLite (Linux/macOS/Windows)
     "tiktoken>=0.8.0,<1.0.0",
     "litellm>=1.83.7",
-    "instructor>=1.9.1,<1.15.3",
+    "instructor>=1.9.1,<2.0.0",
     "filetype>=1.2.0,<2.0.0",
     "aiohttp>=3.13.3,<4.0.0",
     "aiofiles>=23.2.1",