Merge branch 'dev' into pgvector-add-normalization

.github/workflows/clean_stale_pr.yaml

@@ -0,0 +1,24 @@
+name: clean | remove stale PRs
+
+on:
+  # Run this action periodically (daily at 0:00 UTC in this example).
+  schedule:
+    - cron: "0 0 * * *"
+  # Optionally, also run when pull requests are labeled, unlabeled, synchronized, or reopened
+  # to update the stale timer as needed. Uncomment if desired.
+  # pull_request:
+  #   types: [labeled, unlabeled, synchronize, reopened]
+
+jobs:
+  stale:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Mark and Close Stale
+        uses: actions/stale@v6
+        with:
+          # Number of days of inactivity before the pull request is marked stale
+          days-before-stale: 60
+          # Number of days of inactivity after being marked stale before the pull request is closed
+          days-before-close: 7
+          # Comment to post when marking as stale
+          stale-pr-message: "This pull request has been automatically marke

README.md

@@ -217,10 +217,11 @@ Cognee supports a variety of tools and services for different operations:
 
 ## Demo
 
-Check out our demo notebook [here](https://github.com/topoteretes/cognee/blob/main/notebooks/cognee_demo.ipynb)
+Check out our demo notebook [here](https://github.com/topoteretes/cognee/blob/main/notebooks/cognee_demo.ipynb) or watch the Youtube video bellow
 
 
-[<img src="https://i3.ytimg.com/vi/-ARUfIzhzC4/maxresdefault.jpg" width="100%">](https://www.youtube.com/watch?v=BDFt4xVPmro "Learn about cognee: 55")
+[<img src="https://img.youtube.com/vi/fI4hDzguN5k/maxresdefault.jpg" width="100%">](https://www.youtube.com/watch?v=fI4hDzguN5k "Learn about cognee: 55")
+
 
 
 ## Get Started

cognee-mcp/src/__init__.py

@@ -1,6 +1,8 @@
-from .server import mcp
+from .server import main as server_main
 
 
 def main():
     """Main entry point for the package."""
-    mcp.run(transport="stdio")
+    import asyncio
+
+    asyncio.run(server_main())

cognee-mcp/src/client.py

@@ -4,8 +4,8 @@
 
 # Create server parameters for stdio connection
 server_params = StdioServerParameters(
-    command="mcp",  # Executable
-    args=["run", "src/server.py"],  # Optional command line arguments
+    command="uv",  # Executable
+    args=["--directory", ".", "run", "cognee"],  # Optional command line arguments
     env=None,  # Optional environment variables
 )
 
@@ -33,10 +33,15 @@ async def run():
         async with ClientSession(read, write, timedelta(minutes=3)) as session:
             await session.initialize()
 
+            toolResult = await session.list_tools()
+
+            toolResult = await session.call_tool("prune", arguments={})
+
             toolResult = await session.call_tool("cognify", arguments={"text": text})
-            # toolResult = await session.call_tool("search", arguments={"search_query": "AI"})
 
-            print(f"Cognify result: {toolResult}")
+            toolResult = await session.call_tool("search", arguments={"search_query": "AI"})
+
+            print(f"Cognify result: {toolResult.content}")
 
 
 if __name__ == "__main__":

cognee-mcp/src/server.py

@@ -1,16 +1,97 @@
 import os
 import cognee
+import logging
 import importlib.util
+from contextlib import redirect_stderr, redirect_stdout
 
 # from PIL import Image as PILImage
-from mcp.server.fastmcp import FastMCP
+import mcp.types as types
+from mcp.server import Server, NotificationOptions
+from mcp.server.models import InitializationOptions
 from cognee.api.v1.search import SearchType
 from cognee.shared.data_models import KnowledgeGraph
 
-mcp = FastMCP("cognee", timeout=120000)
+mcp = Server("cognee")
+
+logger = logging.getLogger(__name__)
+
+
+@mcp.list_tools()
+async def list_tools() -> list[types.Tool]:
+    return [
+        types.Tool(
+            name="cognify",
+            description="Cognifies text into knowledge graph",
+            inputSchema={
+                "type": "object",
+                "properties": {
+                    "text": {
+                        "type": "string",
+                        "description": "The text to cognify",
+                    },
+                    "graph_model_file": {
+                        "type": "string",
+                        "description": "The path to the graph model file",
+                    },
+                    "graph_model_name": {
+                        "type": "string",
+                        "description": "The name of the graph model",
+                    },
+                },
+                "required": ["text"],
+            },
+        ),
+        types.Tool(
+            name="search",
+            description="Searches for information in knowledge graph",
+            inputSchema={
+                "type": "object",
+                "properties": {
+                    "search_query": {
+                        "type": "string",
+                        "description": "The query to search for",
+                    },
+                },
+                "required": ["search_query"],
+            },
+        ),
+        types.Tool(
+            name="prune",
+            description="Prunes knowledge graph",
+            inputSchema={
+                "type": "object",
+                "properties": {},
+            },
+        ),
+    ]
+
+
+@mcp.call_tool()
+async def call_tools(name: str, arguments: dict) -> list[types.TextContent]:
+    try:
+        with open(os.devnull, "w") as fnull:
+            with redirect_stdout(fnull), redirect_stderr(fnull):
+                if name == "cognify":
+                    await cognify(
+                        text=arguments["text"],
+                        graph_model_file=arguments.get("graph_model_file", None),
+                        graph_model_name=arguments.get("graph_model_name", None),
+                    )
+
+                    return [types.TextContent(type="text", text="Ingested")]
+                elif name == "search":
+                    search_results = await search(arguments["search_query"])
+
+                    return [types.TextContent(type="text", text=search_results)]
+                elif name == "prune":
+                    await prune()
+
+                    return [types.TextContent(type="text", text="Pruned")]
+    except Exception as e:
+        logger.error(f"Error calling tool '{name}': {str(e)}")
+        return [types.TextContent(type="text", text=f"Error calling tool '{name}': {str(e)}")]
 
 
-@mcp.tool()
 async def cognify(text: str, graph_model_file: str = None, graph_model_name: str = None) -> str:
     """Build knowledge graph from the input text"""
     if graph_model_file and graph_model_name:
@@ -25,10 +106,7 @@ async def cognify(text: str, graph_model_file: str = None, graph_model_name: str
     except Exception as e:
         raise ValueError(f"Failed to cognify: {str(e)}")
 
-    return "Ingested"
 
-
-@mcp.tool()
 async def search(search_query: str) -> str:
     """Search the knowledge graph"""
     search_results = await cognee.search(SearchType.INSIGHTS, query_text=search_query)
@@ -38,16 +116,36 @@ async def search(search_query: str) -> str:
     return results
 
 
-@mcp.tool()
 async def prune() -> str:
     """Reset the knowledge graph"""
     await cognee.prune.prune_data()
     await cognee.prune.prune_system(metadata=True)
 
-    return "Pruned"
+
+async def main():
+    try:
+        from mcp.server.stdio import stdio_server
+
+        async with stdio_server() as (read_stream, write_stream):
+            await mcp.run(
+                read_stream=read_stream,
+                write_stream=write_stream,
+                initialization_options=InitializationOptions(
+                    server_name="cognee",
+                    server_version="0.1.0",
+                    capabilities=mcp.get_capabilities(
+                        notification_options=NotificationOptions(),
+                        experimental_capabilities={},
+                    ),
+                ),
+                raise_exceptions=True,
+            )
+
+    except Exception as e:
+        logger.error(f"Server failed to start: {str(e)}", exc_info=True)
+        raise
 
 
-# @mcp.tool()
 # async def visualize() -> Image:
 #     """Visualize the knowledge graph"""
 #     try:
@@ -116,4 +214,6 @@ def load_class(model_file, model_name):
 
 if __name__ == "__main__":
     # Initialize and run the server
-    mcp.run(transport="stdio")
+    import asyncio
+
+    asyncio.run(main())