[{"_1":2,"_490":-5,"_491":-5},"loaderData",{"_3":4,"_24":25,"_318":319},"root",{"_5":6,"_7":8},"toasts",[],"visitor",{"_9":10,"_11":12},"viewportWidth",1280,"visitorSession",{"_13":14,"_16":17,"_18":-5,"_19":20,"_21":22,"_23":-5},"attributes",[15],"bot","id",27182804,"membership","referenceId","019ab33c-32bd-768e-80e6-9ee61bfabc38","unreadReleaseNotesCount",2,"userAccount","routes/_public/mcp/servers/~namespace/~slug/_pages/_layout",{"_26":27,"_75":-5,"_76":40,"_29":77,"_236":237,"_317":219},"availableLocales",[28,40,45,51,57,63,69],{"_29":30,"_36":37,"_38":39},"mcpServer",{"_31":32,"_33":32,"_34":35},"description","Graphiti MCP Server","displayName","readme","# Graphiti MCP Server\n\nGraphiti ist ein Framework zum Erstellen und Abfragen zeitabhängiger Wissensgraphen, das speziell auf KI-Agenten in dynamischen Umgebungen zugeschnitten ist. Im Gegensatz zu herkömmlichen Retrieval-Augmented-Generation-Methoden (RAG) integriert Graphiti kontinuierlich Benutzerinteraktionen, strukturierte und unstrukturierte Unternehmensdaten sowie externe Informationen in einen kohärenten, abfragbaren Graphen. Das Framework unterstützt inkrementelle Datenaktualisierungen, effiziente Abfragen und präzise historische Abfragen ohne vollständige Neuberechnung des Graphen und eignet sich daher für die Entwicklung interaktiver, kontextabhängiger KI-Anwendungen.\n\nDies ist eine experimentelle Serverimplementierung des Model Context Protocol (MCP) für Graphiti. Der MCP-Server stellt die wichtigsten Funktionen von Graphiti über das MCP-Protokoll bereit und ermöglicht KI-Assistenten die Interaktion mit den Wissensgraphenfunktionen von Graphiti.\n\n## Merkmale\n\nDer Graphiti MCP-Server stellt die folgenden wichtigen High-Level-Funktionen von Graphiti bereit:\n\n* **Episodenverwaltung** : Episoden (Text, Nachrichten oder JSON-Daten) hinzufügen, abrufen und löschen\n* **Entity Management** : Suchen und verwalten Sie Entity-Knoten und -Beziehungen im Wissensgraphen\n* **Suchfunktionen** : Suche nach Fakten (Kanten) und Knotenzusammenfassungen mithilfe der semantischen und hybriden Suche\n* **Gruppenverwaltung** : Organisieren und verwalten Sie Gruppen verwandter Daten mit group\\_id-Filterung\n* **Graphenwartung** : Löschen Sie den Graphen und erstellen Sie die Indizes neu\n\n## Schnellstart für Claude Desktop, Cursor und andere Clients\n\n1. Klonen Sie das Graphiti GitHub-Repository\n\n```bash\ngit clone https://github.com/getzep/graphiti.git\n```\n\noder\n\n```bash\ngh repo clone getzep/graphiti\n```\n\nNotieren Sie den vollständigen Pfad zu diesem Verzeichnis.\n\n```\ncd graphiti && pwd\n```\n\n2. Installieren Sie die [Graphiti-Voraussetzungen](#prerequisites) .\n\n3. Konfigurieren Sie Claude, Cursor oder andere MCP-Clients für die Verwendung [von Graphiti mit einem `stdio` Transport](#integrating-with-mcp-clients) . Informationen zum Speicherort der MCP-Konfigurationsdateien finden Sie in der Clientdokumentation.\n\n## Installation\n\n### Voraussetzungen\n\n1. Stellen Sie sicher, dass Sie Python 3.10 oder höher installiert haben.\n2. Eine laufende Neo4j-Datenbank (Version 5.26 oder höher erforderlich)\n3. OpenAI-API-Schlüssel für LLM-Operationen\n\n### Aufstellen\n\n1. Klonen Sie das Repository und navigieren Sie zum Verzeichnis mcp\\_server\n2. Verwenden Sie `uv` , um eine virtuelle Umgebung zu erstellen und Abhängigkeiten zu installieren:\n\n```bash\n# Install uv if you don't have it already\ncurl -LsSf https://astral.sh/uv/install.sh | sh\n\n# Create a virtual environment and install dependencies in one step\nuv sync\n```\n\n## Konfiguration\n\nDer Server verwendet die folgenden Umgebungsvariablen:\n\n* `NEO4J_URI` : URI für die Neo4j-Datenbank (Standard: `bolt://localhost:7687` )\n* `NEO4J_USER` : Neo4j-Benutzername (Standard: `neo4j` )\n* `NEO4J_PASSWORD` : Neo4j-Passwort (Standard: `demodemo` )\n* `OPENAI_API_KEY` : OpenAI-API-Schlüssel (erforderlich für LLM-Operationen)\n* `OPENAI_BASE_URL` : Optionale Basis-URL für die OpenAI-API\n* `MODEL_NAME` : Für LLM-Operationen zu verwendender OpenAI-Modellname.\n* `SMALL_MODEL_NAME` : OpenAI-Modellname zur Verwendung für kleinere LLM-Operationen.\n* `LLM_TEMPERATURE` : Temperatur für LLM-Antworten (0,0–2,0).\n* `AZURE_OPENAI_ENDPOINT` : Optionale Azure OpenAI-Endpunkt-URL\n* `AZURE_OPENAI_DEPLOYMENT_NAME` : Optionaler Azure OpenAI-Bereitstellungsname\n* `AZURE_OPENAI_API_VERSION` : Optionale Azure OpenAI API-Version\n* `AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME` : Optionaler Azure OpenAI-Einbettungsbereitstellungsname\n* `AZURE_OPENAI_EMBEDDING_API_VERSION` : Optionale Azure OpenAI API-Version\n* `AZURE_OPENAI_USE_MANAGED_IDENTITY` : Optionale Verwendung von Azure Managed Identities zur Authentifizierung\n\nSie können diese Variablen in einer `.env` Datei im Projektverzeichnis festlegen.\n\n## Ausführen des Servers\n\nSo führen Sie den Graphiti MCP-Server direkt mit `uv` aus:\n\n```bash\nuv run graphiti_mcp_server.py\n```\n\nMit Optionen:\n\n```bash\nuv run graphiti_mcp_server.py --model gpt-4.1-mini --transport sse\n```\n\nVerfügbare Argumente:\n\n* `--model` : Überschreibt die Umgebungsvariable `MODEL_NAME` .\n* `--small-model` : Überschreibt die Umgebungsvariable `SMALL_MODEL_NAME` .\n* `--temperature` : Überschreibt die Umgebungsvariable `LLM_TEMPERATURE` .\n* `--transport` : Wählen Sie die Transportmethode (sse oder stdio, Standard: sse)\n* `--group-id` : Legt einen Namespace für den Graphen fest (optional). Falls nicht angegeben, wird standardmäßig „default“ verwendet.\n* `--destroy-graph` : Wenn festgelegt, werden beim Start alle Graphiti-Diagramme zerstört.\n* `--use-custom-entities` : Aktiviert die Entitätsextraktion mit den vordefinierten ENTITY\\_TYPES\n\n### Docker-Bereitstellung\n\nDer Graphiti MCP-Server kann mit Docker bereitgestellt werden. Das Dockerfile verwendet `uv` für die Paketverwaltung und gewährleistet so eine konsistente Installation der Abhängigkeiten.\n\n#### Umgebungskonfiguration\n\nBevor Sie das Docker Compose-Setup ausführen, müssen Sie die Umgebungsvariablen konfigurieren. Sie haben zwei Möglichkeiten:\n\n1. **Verwenden einer .env-Datei** (empfohlen):\n\n * Kopieren Sie die bereitgestellte `.env.example` -Datei, um eine `.env` Datei zu erstellen:\n\n ```bash\n cp .env.example .env\n ```\n\n * Bearbeiten Sie die `.env` Datei, um Ihren OpenAI-API-Schlüssel und andere Konfigurationsoptionen festzulegen:\n\n ```\n # Required for LLM operations\n OPENAI_API_KEY=your_openai_api_key_here\n MODEL_NAME=gpt-4.1-mini\n # Optional: OPENAI_BASE_URL only needed for non-standard OpenAI endpoints\n # OPENAI_BASE_URL=https://api.openai.com/v1\n ```\n\n * Das Docker Compose-Setup ist so konfiguriert, dass diese Datei verwendet wird, sofern sie vorhanden ist (optional).\n\n2. **Umgebungsvariablen direkt verwenden** :\n\n * Sie können die Umgebungsvariablen auch beim Ausführen des Docker Compose-Befehls festlegen:\n\n ```bash\n OPENAI_API_KEY=your_key MODEL_NAME=gpt-4.1-mini docker compose up\n ```\n\n#### Neo4j-Konfiguration\n\nDas Docker Compose-Setup umfasst einen Neo4j-Container mit der folgenden Standardkonfiguration:\n\n* Benutzername: `neo4j`\n* Passwort: `demodemo`\n* URI: `bolt://neo4j:7687` (innerhalb des Docker-Netzwerks)\n* Für die Entwicklung optimierte Speichereinstellungen\n\n#### Ausführen mit Docker Compose\n\nStarten Sie die Dienste mit Docker Compose:\n\n```bash\ndocker compose up\n```\n\nOder wenn Sie eine ältere Version von Docker Compose verwenden:\n\n```bash\ndocker-compose up\n```\n\nDadurch werden sowohl die Neo4j-Datenbank als auch der Graphiti MCP-Server gestartet. Das Docker-Setup:\n\n* Verwendet `uv` für die Paketverwaltung und den Betrieb des Servers\n* Installiert Abhängigkeiten aus der Datei `pyproject.toml`\n* Stellt über die Umgebungsvariablen eine Verbindung zum Neo4j-Container her\n* Stellt den Server auf Port 8000 für HTTP-basierten SSE-Transport bereit\n* Beinhaltet einen Healthcheck für Neo4j, um sicherzustellen, dass es vor dem Starten des MCP-Servers voll funktionsfähig ist\n\n## Integration mit MCP-Clients\n\n### Konfiguration\n\nUm den Graphiti MCP-Server mit einem MCP-kompatiblen Client zu verwenden, konfigurieren Sie ihn für die Verbindung mit dem Server:\n\n> \\[!WICHTIG] Sie benötigen den Python-Paketmanager `uv` . Bitte beachten Sie die [`uv` Installationsanweisungen](https://docs.astral.sh/uv/getting-started/installation/) .\n>\n> Stellen Sie sicher, dass Sie den vollständigen Pfad zur `uv` Binärdatei und Ihrem Graphiti-Projektordner festlegen.\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"stdio\",\n \"command\": \"/Users//.local/bin/uv\",\n \"args\": [\n \"run\",\n \"--isolated\",\n \"--directory\",\n \"/Users/>/dev/zep/graphiti/mcp_server\",\n \"--project\",\n \".\",\n \"graphiti_mcp_server.py\",\n \"--transport\",\n \"stdio\"\n ],\n \"env\": {\n \"NEO4J_URI\": \"bolt://localhost:7687\",\n \"NEO4J_USER\": \"neo4j\",\n \"NEO4J_PASSWORD\": \"password\",\n \"OPENAI_API_KEY\": \"sk-XXXXXXXX\",\n \"MODEL_NAME\": \"gpt-4.1-mini\"\n }\n }\n }\n}\n```\n\nFür den SSE-Transport (HTTP-basiert) können Sie diese Konfiguration verwenden:\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"sse\",\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n## Verfügbare Tools\n\nDer Graphiti MCP-Server stellt die folgenden Tools bereit:\n\n* `add_episode` : Fügen Sie dem Wissensgraphen eine Episode hinzu (unterstützt Text-, JSON- und Nachrichtenformate)\n* `search_nodes` : Durchsuchen Sie den Wissensgraphen nach relevanten Knotenzusammenfassungen\n* `search_facts` : Durchsucht den Wissensgraphen nach relevanten Fakten (Kanten zwischen Entitäten)\n* `delete_entity_edge` : Löscht eine Entitätskante aus dem Wissensgraphen\n* `delete_episode` : Löscht eine Episode aus dem Wissensgraphen\n* `get_entity_edge` : Ruft eine Entitätskante über ihre UUID ab.\n* `get_episodes` : Holen Sie sich die neuesten Episoden für eine bestimmte Gruppe\n* `clear_graph` : Alle Daten aus dem Wissensgraphen löschen und Indizes neu erstellen\n* `get_status` : Ruft den Status des Graphiti MCP-Servers und der Neo4j-Verbindung ab\n\n## Arbeiten mit JSON-Daten\n\nDer Graphiti MCP-Server kann strukturierte JSON-Daten über das Tool `add_episode` mit `source=\"json\"` verarbeiten. Dadurch können Sie Entitäten und Beziehungen automatisch aus strukturierten Daten extrahieren:\n\n```\n\nadd_episode(\nname=\"Customer Profile\",\nepisode_body=\"{\\\"company\\\": {\\\"name\\\": \\\"Acme Technologies\\\"}, \\\"products\\\": [{\\\"id\\\": \\\"P001\\\", \\\"name\\\": \\\"CloudSync\\\"}, {\\\"id\\\": \\\"P002\\\", \\\"name\\\": \\\"DataMiner\\\"}]}\",\nsource=\"json\",\nsource_description=\"CRM data\"\n)\n\n```\n\n## Integration mit der Cursor-IDE\n\nUm den Graphiti MCP-Server in die Cursor IDE zu integrieren, gehen Sie folgendermaßen vor:\n\n1. Führen Sie den Graphiti MCP-Server mithilfe des SSE-Transports aus:\n\n```bash\npython graphiti_mcp_server.py --transport sse --use-custom-entities --group-id \n```\n\nHinweis: Geben Sie eine `group_id` an, um Graphdaten zu benennen. Wenn Sie keine `group_id` angeben, verwendet der Server „default“ als Gruppen-ID.\n\noder\n\n```bash\ndocker compose up\n```\n\n2. Konfigurieren Sie Cursor für die Verbindung mit dem Graphiti MCP-Server.\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n3. Fügen Sie die Graphiti-Regeln zu den Benutzerregeln von Cursor hinzu. Weitere Informationen finden Sie in [cursor\\_rules.md](cursor_rules.md) .\n\n4. Starten Sie eine Agentensitzung in Cursor.\n\nDie Integration ermöglicht es KI-Assistenten in Cursor, durch die Wissensgraphenfunktionen von Graphiti einen dauerhaften Speicher aufrechtzuerhalten.\n\n## Integration mit Claude Desktop (Docker MCP Server)\n\nDer Graphiti MCP Server-Container verwendet den SSE MCP-Transport. Claude Desktop unterstützt SSE nicht nativ, daher müssen Sie ein Gateway wie `mcp-remote` verwenden.\n\n1. **Führen Sie den Graphiti MCP-Server mit SSE-Transport aus** :\n\n ```bash\n docker compose up\n ```\n\n2. **(Optional) Installieren Sie `mcp-remote` global** : Wenn Sie `mcp-remote` lieber global installieren möchten oder Probleme beim Abrufen des Pakets durch `npx` auftreten, können Sie es global installieren. Andernfalls übernimmt `npx` (wird im nächsten Schritt verwendet) die Installation für Sie.\n\n ```bash\n npm install -g mcp-remote\n ```\n\n3. **Konfigurieren Sie Claude Desktop** : Öffnen Sie Ihre Claude Desktop-Konfigurationsdatei (normalerweise `claude_desktop_config.json` ) und fügen Sie den Abschnitt `mcpServers` wie folgt hinzu oder ändern Sie ihn:\n\n ```json\n {\n \"mcpServers\": {\n \"graphiti-memory\": {\n // You can choose a different name if you prefer\n \"command\": \"npx\", // Or the full path to mcp-remote if npx is not in your PATH\n \"args\": [\n \"mcp-remote\",\n \"http://localhost:8000/sse\" // Ensure this matches your Graphiti server's SSE endpoint\n ]\n }\n }\n }\n ```\n\n Wenn Sie bereits einen `mcpServers` Eintrag haben, fügen Sie `graphiti-memory` (oder Ihren gewählten Namen) als neuen Schlüssel hinzu.\n\n4. **Starten Sie Claude Desktop neu,** damit die Änderungen wirksam werden.\n\n## Anforderungen\n\n* Python 3.10 oder höher\n* Neo4j-Datenbank (Version 5.26 oder höher erforderlich)\n* OpenAI-API-Schlüssel (für LLM-Operationen und -Einbettungen)\n* MCP-kompatibler Client\n\n## Lizenz\n\nDieses Projekt unterliegt derselben Lizenz wie das übergeordnete Graphiti-Projekt.","name","German","tag","de-DE",{"_29":41,"_36":43,"_38":44},{"_31":32,"_33":32,"_34":42},"# Graphiti MCP Server\n\nGraphiti is a framework for building and querying temporally-aware knowledge graphs, specifically tailored for AI agents\noperating in dynamic environments. Unlike traditional retrieval-augmented generation (RAG) methods, Graphiti\ncontinuously integrates user interactions, structured and unstructured enterprise data, and external information into a\ncoherent, queryable graph. The framework supports incremental data updates, efficient retrieval, and precise historical\nqueries without requiring complete graph recomputation, making it suitable for developing interactive, context-aware AI\napplications.\n\nThis is an experimental Model Context Protocol (MCP) server implementation for Graphiti. The MCP server exposes\nGraphiti's key functionality through the MCP protocol, allowing AI assistants to interact with Graphiti's knowledge\ngraph capabilities.\n\n## Features\n\nThe Graphiti MCP server exposes the following key high-level functions of Graphiti:\n\n- **Episode Management**: Add, retrieve, and delete episodes (text, messages, or JSON data)\n- **Entity Management**: Search and manage entity nodes and relationships in the knowledge graph\n- **Search Capabilities**: Search for facts (edges) and node summaries using semantic and hybrid search\n- **Group Management**: Organize and manage groups of related data with group_id filtering\n- **Graph Maintenance**: Clear the graph and rebuild indices\n\n## Quick Start for Claude Desktop, Cursor, and other clients\n\n1. Clone the Graphiti GitHub repo\n\n```bash\ngit clone https://github.com/getzep/graphiti.git\n```\n\nor\n\n```bash\ngh repo clone getzep/graphiti\n```\n\nNote the full path to this directory.\n\n```\ncd graphiti && pwd\n```\n\n2. Install the [Graphiti prerequisites](#prerequisites).\n\n3. Configure Claude, Cursor, or other MCP client to use [Graphiti with a `stdio` transport](#integrating-with-mcp-clients). See the client documentation on where to find their MCP configuration files.\n\n## Installation\n\n### Prerequisites\n\n1. Ensure you have Python 3.10 or higher installed.\n2. A running Neo4j database (version 5.26 or later required)\n3. OpenAI API key for LLM operations\n\n### Setup\n\n1. Clone the repository and navigate to the mcp_server directory\n2. Use `uv` to create a virtual environment and install dependencies:\n\n```bash\n# Install uv if you don't have it already\ncurl -LsSf https://astral.sh/uv/install.sh | sh\n\n# Create a virtual environment and install dependencies in one step\nuv sync\n```\n\n## Configuration\n\nThe server uses the following environment variables:\n\n- `NEO4J_URI`: URI for the Neo4j database (default: `bolt://localhost:7687`)\n- `NEO4J_USER`: Neo4j username (default: `neo4j`)\n- `NEO4J_PASSWORD`: Neo4j password (default: `demodemo`)\n- `OPENAI_API_KEY`: OpenAI API key (required for LLM operations)\n- `OPENAI_BASE_URL`: Optional base URL for OpenAI API\n- `MODEL_NAME`: OpenAI model name to use for LLM operations.\n- `SMALL_MODEL_NAME`: OpenAI model name to use for smaller LLM operations.\n- `LLM_TEMPERATURE`: Temperature for LLM responses (0.0-2.0).\n- `AZURE_OPENAI_ENDPOINT`: Optional Azure OpenAI endpoint URL\n- `AZURE_OPENAI_DEPLOYMENT_NAME`: Optional Azure OpenAI deployment name\n- `AZURE_OPENAI_API_VERSION`: Optional Azure OpenAI API version\n- `AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME`: Optional Azure OpenAI embedding deployment name\n- `AZURE_OPENAI_EMBEDDING_API_VERSION`: Optional Azure OpenAI API version\n- `AZURE_OPENAI_USE_MANAGED_IDENTITY`: Optional use Azure Managed Identities for authentication\n\nYou can set these variables in a `.env` file in the project directory.\n\n## Running the Server\n\nTo run the Graphiti MCP server directly using `uv`:\n\n```bash\nuv run graphiti_mcp_server.py\n```\n\nWith options:\n\n```bash\nuv run graphiti_mcp_server.py --model gpt-4.1-mini --transport sse\n```\n\nAvailable arguments:\n\n- `--model`: Overrides the `MODEL_NAME` environment variable.\n- `--small-model`: Overrides the `SMALL_MODEL_NAME` environment variable.\n- `--temperature`: Overrides the `LLM_TEMPERATURE` environment variable.\n- `--transport`: Choose the transport method (sse or stdio, default: sse)\n- `--group-id`: Set a namespace for the graph (optional). If not provided, defaults to \"default\".\n- `--destroy-graph`: If set, destroys all Graphiti graphs on startup.\n- `--use-custom-entities`: Enable entity extraction using the predefined ENTITY_TYPES\n\n### Docker Deployment\n\nThe Graphiti MCP server can be deployed using Docker. The Dockerfile uses `uv` for package management, ensuring\nconsistent dependency installation.\n\n#### Environment Configuration\n\nBefore running the Docker Compose setup, you need to configure the environment variables. You have two options:\n\n1. **Using a .env file** (recommended):\n\n - Copy the provided `.env.example` file to create a `.env` file:\n ```bash\n cp .env.example .env\n ```\n - Edit the `.env` file to set your OpenAI API key and other configuration options:\n ```\n # Required for LLM operations\n OPENAI_API_KEY=your_openai_api_key_here\n MODEL_NAME=gpt-4.1-mini\n # Optional: OPENAI_BASE_URL only needed for non-standard OpenAI endpoints\n # OPENAI_BASE_URL=https://api.openai.com/v1\n ```\n - The Docker Compose setup is configured to use this file if it exists (it's optional)\n\n2. **Using environment variables directly**:\n - You can also set the environment variables when running the Docker Compose command:\n ```bash\n OPENAI_API_KEY=your_key MODEL_NAME=gpt-4.1-mini docker compose up\n ```\n\n#### Neo4j Configuration\n\nThe Docker Compose setup includes a Neo4j container with the following default configuration:\n\n- Username: `neo4j`\n- Password: `demodemo`\n- URI: `bolt://neo4j:7687` (from within the Docker network)\n- Memory settings optimized for development use\n\n#### Running with Docker Compose\n\nStart the services using Docker Compose:\n\n```bash\ndocker compose up\n```\n\nOr if you're using an older version of Docker Compose:\n\n```bash\ndocker-compose up\n```\n\nThis will start both the Neo4j database and the Graphiti MCP server. The Docker setup:\n\n- Uses `uv` for package management and running the server\n- Installs dependencies from the `pyproject.toml` file\n- Connects to the Neo4j container using the environment variables\n- Exposes the server on port 8000 for HTTP-based SSE transport\n- Includes a healthcheck for Neo4j to ensure it's fully operational before starting the MCP server\n\n## Integrating with MCP Clients\n\n### Configuration\n\nTo use the Graphiti MCP server with an MCP-compatible client, configure it to connect to the server:\n\n> [!IMPORTANT]\n> You will need the Python package manager, `uv` installed. Please refer to the [`uv` install instructions](https://docs.astral.sh/uv/getting-started/installation/).\n>\n> Ensure that you set the full path to the `uv` binary and your Graphiti project folder.\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"stdio\",\n \"command\": \"/Users//.local/bin/uv\",\n \"args\": [\n \"run\",\n \"--isolated\",\n \"--directory\",\n \"/Users/>/dev/zep/graphiti/mcp_server\",\n \"--project\",\n \".\",\n \"graphiti_mcp_server.py\",\n \"--transport\",\n \"stdio\"\n ],\n \"env\": {\n \"NEO4J_URI\": \"bolt://localhost:7687\",\n \"NEO4J_USER\": \"neo4j\",\n \"NEO4J_PASSWORD\": \"password\",\n \"OPENAI_API_KEY\": \"sk-XXXXXXXX\",\n \"MODEL_NAME\": \"gpt-4.1-mini\"\n }\n }\n }\n}\n```\n\nFor SSE transport (HTTP-based), you can use this configuration:\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"sse\",\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n## Available Tools\n\nThe Graphiti MCP server exposes the following tools:\n\n- `add_episode`: Add an episode to the knowledge graph (supports text, JSON, and message formats)\n- `search_nodes`: Search the knowledge graph for relevant node summaries\n- `search_facts`: Search the knowledge graph for relevant facts (edges between entities)\n- `delete_entity_edge`: Delete an entity edge from the knowledge graph\n- `delete_episode`: Delete an episode from the knowledge graph\n- `get_entity_edge`: Get an entity edge by its UUID\n- `get_episodes`: Get the most recent episodes for a specific group\n- `clear_graph`: Clear all data from the knowledge graph and rebuild indices\n- `get_status`: Get the status of the Graphiti MCP server and Neo4j connection\n\n## Working with JSON Data\n\nThe Graphiti MCP server can process structured JSON data through the `add_episode` tool with `source=\"json\"`. This\nallows you to automatically extract entities and relationships from structured data:\n\n```\n\nadd_episode(\nname=\"Customer Profile\",\nepisode_body=\"{\\\"company\\\": {\\\"name\\\": \\\"Acme Technologies\\\"}, \\\"products\\\": [{\\\"id\\\": \\\"P001\\\", \\\"name\\\": \\\"CloudSync\\\"}, {\\\"id\\\": \\\"P002\\\", \\\"name\\\": \\\"DataMiner\\\"}]}\",\nsource=\"json\",\nsource_description=\"CRM data\"\n)\n\n```\n\n## Integrating with the Cursor IDE\n\nTo integrate the Graphiti MCP Server with the Cursor IDE, follow these steps:\n\n1. Run the Graphiti MCP server using the SSE transport:\n\n```bash\npython graphiti_mcp_server.py --transport sse --use-custom-entities --group-id \n```\n\nHint: specify a `group_id` to namespace graph data. If you do not specify a `group_id`, the server will use \"default\" as the group_id.\n\nor\n\n```bash\ndocker compose up\n```\n\n2. Configure Cursor to connect to the Graphiti MCP server.\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n3. Add the Graphiti rules to Cursor's User Rules. See [cursor_rules.md](cursor_rules.md) for details.\n\n4. Kick off an agent session in Cursor.\n\nThe integration enables AI assistants in Cursor to maintain persistent memory through Graphiti's knowledge graph\ncapabilities.\n\n## Integrating with Claude Desktop (Docker MCP Server)\n\nThe Graphiti MCP Server container uses the SSE MCP transport. Claude Desktop does not natively support SSE, so you'll need to use a gateway like `mcp-remote`.\n\n1. **Run the Graphiti MCP server using SSE transport**:\n\n ```bash\n docker compose up\n ```\n\n2. **(Optional) Install `mcp-remote` globally**:\n If you prefer to have `mcp-remote` installed globally, or if you encounter issues with `npx` fetching the package, you can install it globally. Otherwise, `npx` (used in the next step) will handle it for you.\n\n ```bash\n npm install -g mcp-remote\n ```\n\n3. **Configure Claude Desktop**:\n Open your Claude Desktop configuration file (usually `claude_desktop_config.json`) and add or modify the `mcpServers` section as follows:\n\n ```json\n {\n \"mcpServers\": {\n \"graphiti-memory\": {\n // You can choose a different name if you prefer\n \"command\": \"npx\", // Or the full path to mcp-remote if npx is not in your PATH\n \"args\": [\n \"mcp-remote\",\n \"http://localhost:8000/sse\" // Ensure this matches your Graphiti server's SSE endpoint\n ]\n }\n }\n }\n ```\n\n If you already have an `mcpServers` entry, add `graphiti-memory` (or your chosen name) as a new key within it.\n\n4. **Restart Claude Desktop** for the changes to take effect.\n\n## Requirements\n\n- Python 3.10 or higher\n- Neo4j database (version 5.26 or later required)\n- OpenAI API key (for LLM operations and embeddings)\n- MCP-compatible client\n\n## License\n\nThis project is licensed under the same license as the parent Graphiti project.\n","English","en-US",{"_29":46,"_36":49,"_38":50},{"_31":47,"_33":47,"_34":48},"Servidor Graphiti MCP","# Servidor Graphiti MCP\n\nGraphiti es un marco para crear y consultar grafos de conocimiento con conciencia temporal, diseñado específicamente para agentes de IA que operan en entornos dinámicos. A diferencia de los métodos tradicionales de recuperación-generación aumentada (RAG), Graphiti integra continuamente las interacciones del usuario, datos empresariales estructurados y no estructurados, e información externa en un grafo coherente y consultable. El marco admite actualizaciones incrementales de datos, recuperación eficiente y consultas históricas precisas sin necesidad de recálculo completo del grafo, lo que lo hace ideal para el desarrollo de aplicaciones de IA interactivas y contextuales.\n\nEsta es una implementación experimental del servidor del Protocolo de Contexto de Modelo (MCP) para Graphiti. El servidor MCP expone la funcionalidad clave de Graphiti a través del protocolo MCP, lo que permite a los asistentes de IA interactuar con las capacidades del grafo de conocimiento de Graphiti.\n\n## Características\n\nEl servidor Graphiti MCP expone las siguientes funciones clave de alto nivel de Graphiti:\n\n* **Gestión de episodios** : agregue, recupere y elimine episodios (texto, mensajes o datos JSON)\n* **Gestión de entidades** : busque y administre nodos de entidades y relaciones en el gráfico de conocimiento\n* **Capacidades de búsqueda** : Busque hechos (bordes) y resúmenes de nodos mediante búsqueda semántica e híbrida\n* **Gestión de grupos** : organice y administre grupos de datos relacionados con el filtrado group\\_id\n* **Mantenimiento de gráficos** : Limpiar el gráfico y reconstruir los índices\n\n## Inicio rápido para Claude Desktop, Cursor y otros clientes\n\n1. Clonar el repositorio de Graphiti en GitHub\n\n```bash\ngit clone https://github.com/getzep/graphiti.git\n```\n\no\n\n```bash\ngh repo clone getzep/graphiti\n```\n\nAnote la ruta completa a este directorio.\n\n```\ncd graphiti && pwd\n```\n\n2. Instalar los [requisitos previos de Graphiti](#prerequisites) .\n\n3. Configure Claude, Cursor u otro cliente MCP para usar [Graphiti con un transporte `stdio`](#integrating-with-mcp-clients) . Consulte la documentación del cliente para saber dónde encontrar sus archivos de configuración de MCP.\n\n## Instalación\n\n### Prerrequisitos\n\n1. Asegúrese de tener instalado Python 3.10 o superior.\n2. Una base de datos Neo4j en ejecución (se requiere la versión 5.26 o posterior)\n3. Clave API de OpenAI para operaciones LLM\n\n### Configuración\n\n1. Clona el repositorio y navega al directorio mcp\\_server\n2. Utilice `uv` para crear un entorno virtual e instalar dependencias:\n\n```bash\n# Install uv if you don't have it already\ncurl -LsSf https://astral.sh/uv/install.sh | sh\n\n# Create a virtual environment and install dependencies in one step\nuv sync\n```\n\n## Configuración\n\nEl servidor utiliza las siguientes variables de entorno:\n\n* `NEO4J_URI` : URI para la base de datos Neo4j (predeterminado: `bolt://localhost:7687` )\n* `NEO4J_USER` : nombre de usuario de Neo4j (predeterminado: `neo4j` )\n* `NEO4J_PASSWORD` : Contraseña de Neo4j (predeterminada: `demodemo` )\n* `OPENAI_API_KEY` : Clave API de OpenAI (necesaria para operaciones LLM)\n* `OPENAI_BASE_URL` : URL base opcional para la API OpenAI\n* `MODEL_NAME` : Nombre del modelo OpenAI que se utilizará para las operaciones LLM.\n* `SMALL_MODEL_NAME` : Nombre del modelo OpenAI que se utilizará para operaciones LLM más pequeñas.\n* `LLM_TEMPERATURE` : Temperatura para las respuestas LLM (0,0-2,0).\n* `AZURE_OPENAI_ENDPOINT` : URL del punto de conexión de Azure OpenAI opcional\n* `AZURE_OPENAI_DEPLOYMENT_NAME` : Nombre de implementación de Azure OpenAI opcional\n* `AZURE_OPENAI_API_VERSION` : Versión opcional de la API de Azure OpenAI\n* `AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME` : Nombre de implementación de incrustación de Azure OpenAI opcional\n* `AZURE_OPENAI_EMBEDDING_API_VERSION` : Versión opcional de la API de Azure OpenAI\n* `AZURE_OPENAI_USE_MANAGED_IDENTITY` : Uso opcional de identidades administradas de Azure para la autenticación\n\nPuede configurar estas variables en un archivo `.env` en el directorio del proyecto.\n\n## Ejecución del servidor\n\nPara ejecutar el servidor Graphiti MCP directamente usando `uv` :\n\n```bash\nuv run graphiti_mcp_server.py\n```\n\nCon opciones:\n\n```bash\nuv run graphiti_mcp_server.py --model gpt-4.1-mini --transport sse\n```\n\nArgumentos disponibles:\n\n* `--model` : anula la variable de entorno `MODEL_NAME` .\n* `--small-model` : anula la variable de entorno `SMALL_MODEL_NAME` .\n* `--temperature` : anula la variable de entorno `LLM_TEMPERATURE` .\n* `--transport` : elige el método de transporte (sse o stdio, predeterminado: sse)\n* `--group-id` : Establece un espacio de nombres para el gráfico (opcional). Si no se proporciona, el valor predeterminado es \"default\".\n* `--destroy-graph` : si se configura, destruye todos los gráficos de Graphiti al iniciarse.\n* `--use-custom-entities` : Habilitar la extracción de entidades utilizando los ENTITY\\_TYPES predefinidos\n\n### Implementación de Docker\n\nEl servidor Graphiti MCP se puede implementar con Docker. El Dockerfile utiliza `uv` para la gestión de paquetes, lo que garantiza una instalación consistente de las dependencias.\n\n#### Configuración del entorno\n\nAntes de ejecutar la configuración de Docker Compose, debe configurar las variables de entorno. Tiene dos opciones:\n\n1. **Usando un archivo .env** (recomendado):\n\n * Copie el archivo `.env.example` proporcionado para crear un archivo `.env` :\n\n ```bash\n cp .env.example .env\n ```\n\n * Edite el archivo `.env` para configurar su clave API de OpenAI y otras opciones de configuración:\n\n ```\n # Required for LLM operations\n OPENAI_API_KEY=your_openai_api_key_here\n MODEL_NAME=gpt-4.1-mini\n # Optional: OPENAI_BASE_URL only needed for non-standard OpenAI endpoints\n # OPENAI_BASE_URL=https://api.openai.com/v1\n ```\n\n * La configuración de Docker Compose está configurada para usar este archivo si existe (es opcional)\n\n2. **Usando variables de entorno directamente** :\n\n * También puedes configurar las variables de entorno al ejecutar el comando Docker Compose:\n\n ```bash\n OPENAI_API_KEY=your_key MODEL_NAME=gpt-4.1-mini docker compose up\n ```\n\n#### Configuración de Neo4j\n\nLa configuración de Docker Compose incluye un contenedor Neo4j con la siguiente configuración predeterminada:\n\n* Nombre de usuario: `neo4j`\n* Contraseña: `demodemo`\n* URI: `bolt://neo4j:7687` (desde dentro de la red Docker)\n* Configuraciones de memoria optimizadas para uso en desarrollo\n\n#### Ejecutar con Docker Compose\n\nInicie los servicios usando Docker Compose:\n\n```bash\ndocker compose up\n```\n\nO si estás usando una versión anterior de Docker Compose:\n\n```bash\ndocker-compose up\n```\n\nEsto iniciará la base de datos Neo4j y el servidor Graphiti MCP. Configuración de Docker:\n\n* Utiliza `uv` para la gestión de paquetes y la ejecución del servidor.\n* Instala dependencias del archivo `pyproject.toml`\n* Se conecta al contenedor Neo4j mediante las variables de entorno\n* Expone el servidor en el puerto 8000 para el transporte SSE basado en HTTP\n* Incluye una comprobación del estado de Neo4j para garantizar que esté completamente operativo antes de iniciar el servidor MCP\n\n## Integración con clientes MCP\n\n### Configuración\n\nPara utilizar el servidor Graphiti MCP con un cliente compatible con MCP, configúrelo para que se conecte al servidor:\n\n> \\[!IMPORTANTE] Necesitará tener instalado el gestor de paquetes de Python, `uv` . Consulte las [instrucciones de instalación `uv`](https://docs.astral.sh/uv/getting-started/installation/) .\n>\n> Asegúrese de configurar la ruta completa al binario `uv` y a la carpeta de su proyecto Graphiti.\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"stdio\",\n \"command\": \"/Users//.local/bin/uv\",\n \"args\": [\n \"run\",\n \"--isolated\",\n \"--directory\",\n \"/Users/>/dev/zep/graphiti/mcp_server\",\n \"--project\",\n \".\",\n \"graphiti_mcp_server.py\",\n \"--transport\",\n \"stdio\"\n ],\n \"env\": {\n \"NEO4J_URI\": \"bolt://localhost:7687\",\n \"NEO4J_USER\": \"neo4j\",\n \"NEO4J_PASSWORD\": \"password\",\n \"OPENAI_API_KEY\": \"sk-XXXXXXXX\",\n \"MODEL_NAME\": \"gpt-4.1-mini\"\n }\n }\n }\n}\n```\n\nPara el transporte SSE (basado en HTTP), puede utilizar esta configuración:\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"sse\",\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n## Herramientas disponibles\n\nEl servidor Graphiti MCP expone las siguientes herramientas:\n\n* `add_episode` : agrega un episodio al gráfico de conocimiento (admite formatos de texto, JSON y mensajes)\n* `search_nodes` : Busca en el gráfico de conocimiento resúmenes de nodos relevantes\n* `search_facts` : Busca en el gráfico de conocimiento hechos relevantes (bordes entre entidades)\n* `delete_entity_edge` : elimina un borde de entidad del gráfico de conocimiento\n* `delete_episode` : Eliminar un episodio del gráfico de conocimiento\n* `get_entity_edge` : Obtener el borde de una entidad por su UUID\n* `get_episodes` : Obtener los episodios más recientes de un grupo específico\n* `clear_graph` : Borra todos los datos del gráfico de conocimiento y reconstruye los índices\n* `get_status` : obtiene el estado del servidor Graphiti MCP y la conexión Neo4j\n\n## Trabajar con datos JSON\n\nEl servidor Graphiti MCP puede procesar datos JSON estructurados mediante la herramienta `add_episode` con `source=\"json\"` . Esto permite extraer automáticamente entidades y relaciones de datos estructurados:\n\n```\n\nadd_episode(\nname=\"Customer Profile\",\nepisode_body=\"{\\\"company\\\": {\\\"name\\\": \\\"Acme Technologies\\\"}, \\\"products\\\": [{\\\"id\\\": \\\"P001\\\", \\\"name\\\": \\\"CloudSync\\\"}, {\\\"id\\\": \\\"P002\\\", \\\"name\\\": \\\"DataMiner\\\"}]}\",\nsource=\"json\",\nsource_description=\"CRM data\"\n)\n\n```\n\n## Integración con el IDE de Cursor\n\nPara integrar el servidor Graphiti MCP con Cursor IDE, siga estos pasos:\n\n1. Ejecute el servidor Graphiti MCP utilizando el transporte SSE:\n\n```bash\npython graphiti_mcp_server.py --transport sse --use-custom-entities --group-id \n```\n\nConsejo: especifique un `group_id` para los datos del gráfico del espacio de nombres. Si no lo `group_id` , el servidor usará \"default\" como group\\_id.\n\no\n\n```bash\ndocker compose up\n```\n\n2. Configure Cursor para conectarse al servidor Graphiti MCP.\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n3. Añade las reglas de Graphiti a las reglas de usuario de Cursor. Consulta [cursor\\_rules.md](cursor_rules.md) para más detalles.\n\n4. Inicie una sesión de agente en Cursor.\n\nLa integración permite que los asistentes de IA en Cursor mantengan una memoria persistente a través de las capacidades de gráficos de conocimiento de Graphiti.\n\n## Integración con Claude Desktop (Docker MCP Server)\n\nEl contenedor del servidor Graphiti MCP utiliza el transporte SSE MCP. Claude Desktop no es compatible con SSE de forma nativa, por lo que deberá usar una puerta de enlace como `mcp-remote` .\n\n1. **Ejecute el servidor Graphiti MCP utilizando el transporte SSE** :\n\n ```bash\n docker compose up\n ```\n\n2. **(Opcional) Instalar `mcp-remote` globalmente** : Si prefiere tener `mcp-remote` instalado globalmente o si tiene problemas con `npx` al obtener el paquete, puede instalarlo globalmente. De lo contrario, `npx` (usado en el siguiente paso) se encargará de ello.\n\n ```bash\n npm install -g mcp-remote\n ```\n\n3. **Configurar Claude Desktop** : abra el archivo de configuración de Claude Desktop (normalmente `claude_desktop_config.json` ) y agregue o modifique la sección `mcpServers` de la siguiente manera:\n\n ```json\n {\n \"mcpServers\": {\n \"graphiti-memory\": {\n // You can choose a different name if you prefer\n \"command\": \"npx\", // Or the full path to mcp-remote if npx is not in your PATH\n \"args\": [\n \"mcp-remote\",\n \"http://localhost:8000/sse\" // Ensure this matches your Graphiti server's SSE endpoint\n ]\n }\n }\n }\n ```\n\n Si ya tiene una entrada `mcpServers` , agregue `graphiti-memory` (o el nombre elegido) como una nueva clave dentro de ella.\n\n4. **Reinicie Claude Desktop** para que los cambios surtan efecto.\n\n## Requisitos\n\n* Python 3.10 o superior\n* Base de datos Neo4j (se requiere versión 5.26 o posterior)\n* Clave API de OpenAI (para operaciones e incrustaciones LLM)\n* Cliente compatible con MCP\n\n## Licencia\n\nEste proyecto tiene la misma licencia que el proyecto principal Graphiti.","Spanish","es-ES",{"_29":52,"_36":55,"_38":56},{"_31":53,"_33":53,"_34":54},"Graphiti MCP サーバー","# Graphiti MCP サーバー\n\nGraphitiは、時間的変化を考慮したナレッジグラフの構築とクエリのためのフレームワークであり、動的な環境で動作するAIエージェント向けに特化されています。従来の検索拡張型生成（RAG）手法とは異なり、Graphitiはユーザーインタラクション、構造化および非構造化エンタープライズデータ、そして外部情報を、一貫性がありクエリ可能なグラフに継続的に統合します。このフレームワークは、グラフ全体の再計算を必要とせずに、増分データ更新、効率的な検索、そして正確な履歴クエリをサポートしているため、インタラクティブでコンテキストアウェアなAIアプリケーションの開発に最適です。\n\nこれはGraphiti向けの実験的なモデルコンテキストプロトコル（MCP）サーバー実装です。MCPサーバーはGraphitiの主要機能をMCPプロトコルを通じて公開し、AIアシスタントがGraphitiのナレッジグラフ機能を利用できるようにします。\n\n## 特徴\n\nGraphiti MCP サーバーは、Graphiti の次の主要な高レベル機能を公開します。\n\n* **エピソード管理**: エピソード (テキスト、メッセージ、または JSON データ) の追加、取得、削除\n* **エンティティ管理**: ナレッジグラフ内のエンティティノードと関係を検索および管理します。\n* **検索機能**: セマンティック検索とハイブリッド検索を使用してファクト（エッジ）とノードサマリーを検索します\n* **グループ管理**: group\\_id フィルタリングを使用して関連データのグループを整理および管理します。\n* **グラフのメンテナンス**: グラフをクリアしてインデックスを再構築します\n\n## Claude Desktop、Cursor、その他のクライアントのクイックスタート\n\n1. Graphiti GitHubリポジトリをクローンする\n\n```bash\ngit clone https://github.com/getzep/graphiti.git\n```\n\nまたは\n\n```bash\ngh repo clone getzep/graphiti\n```\n\nこのディレクトリへの完全なパスをメモしてください。\n\n```\ncd graphiti && pwd\n```\n\n2. [Graphiti の前提条件](#prerequisites)をインストールします。\n\n3. Claude、Cursor、またはその他のMCPクライアント[を`stdio`トランスポートでGraphitiを](#integrating-with-mcp-clients)使用するように設定します。MCP設定ファイルの場所については、クライアントのドキュメントを参照してください。\n\n## インストール\n\n### 前提条件\n\n1. Python 3.10 以降がインストールされていることを確認してください。\n2. 実行中の Neo4j データベース (バージョン 5.26 以降が必要)\n3. LLM操作用のOpenAI APIキー\n\n### 設定\n\n1. リポジトリをクローンし、mcp\\_serverディレクトリに移動します。\n2. `uv`を使用して仮想環境を作成し、依存関係をインストールします。\n\n```bash\n# Install uv if you don't have it already\ncurl -LsSf https://astral.sh/uv/install.sh | sh\n\n# Create a virtual environment and install dependencies in one step\nuv sync\n```\n\n## 構成\n\nサーバーは次の環境変数を使用します。\n\n* `NEO4J_URI` : Neo4j データベースの URI (デフォルト: `bolt://localhost:7687` )\n* `NEO4J_USER` : Neo4j ユーザー名 (デフォルト: `neo4j` )\n* `NEO4J_PASSWORD` : Neo4j パスワード (デフォルト: `demodemo` )\n* `OPENAI_API_KEY` : OpenAI API キー (LLM 操作に必要)\n* `OPENAI_BASE_URL` : OpenAI API のオプションのベース URL\n* `MODEL_NAME` : LLM 操作に使用する OpenAI モデル名。\n* `SMALL_MODEL_NAME` : 小規模な LLM 操作に使用する OpenAI モデル名。\n* `LLM_TEMPERATURE` : LLM 応答の温度 (0.0-2.0)。\n* `AZURE_OPENAI_ENDPOINT` : オプションの Azure OpenAI エンドポイント URL\n* `AZURE_OPENAI_DEPLOYMENT_NAME` : オプションの Azure OpenAI デプロイメント名\n* `AZURE_OPENAI_API_VERSION` : オプションの Azure OpenAI API バージョン\n* `AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME` : オプションの Azure OpenAI 埋め込みデプロイメント名\n* `AZURE_OPENAI_EMBEDDING_API_VERSION` : オプションの Azure OpenAI API バージョン\n* `AZURE_OPENAI_USE_MANAGED_IDENTITY` : 認証に Azure マネージド ID を使用する (オプション)\n\nこれらの変数は、プロジェクトディレクトリ内の`.env`ファイルで設定できます。\n\n## サーバーの実行\n\n`uv`を使用して Graphiti MCP サーバーを直接実行するには:\n\n```bash\nuv run graphiti_mcp_server.py\n```\n\nオプション付き:\n\n```bash\nuv run graphiti_mcp_server.py --model gpt-4.1-mini --transport sse\n```\n\n使用可能な引数:\n\n* `--model` : `MODEL_NAME`環境変数を上書きします。\n* `--small-model` : `SMALL_MODEL_NAME`環境変数を上書きします。\n* `--temperature` : `LLM_TEMPERATURE`環境変数を上書きします。\n* `--transport` : 転送方法を選択します (sse または stdio、デフォルト: sse)\n* `--group-id` : グラフの名前空間を設定します（オプション）。指定しない場合は、デフォルトで「default」になります。\n* `--destroy-graph` : 設定されている場合、起動時にすべての Graphiti グラフを破棄します。\n* `--use-custom-entities` : 定義済みのENTITY\\_TYPESを使用してエンティティ抽出を有効にする\n\n### Docker デプロイメント\n\nGraphiti MCPサーバーはDockerを使用してデプロイできます。Dockerfileはパッケージ管理に`uv`を使用し、依存関係のインストールの一貫性を確保します。\n\n#### 環境設定\n\nDocker Compose のセットアップを実行する前に、環境変数を設定する必要があります。2 つのオプションがあります。\n\n1. **.env ファイルの使用**(推奨):\n\n * 提供されている`.env.example`ファイルをコピーして`.env`ファイルを作成します。\n\n ```bash\n cp .env.example .env\n ```\n\n * `.env`ファイルを編集して、OpenAI API キーやその他の構成オプションを設定します。\n\n ```\n # Required for LLM operations\n OPENAI_API_KEY=your_openai_api_key_here\n MODEL_NAME=gpt-4.1-mini\n # Optional: OPENAI_BASE_URL only needed for non-standard OpenAI endpoints\n # OPENAI_BASE_URL=https://api.openai.com/v1\n ```\n\n * Docker Compose セットアップは、このファイルが存在する場合はそれを使用するように構成されています (オプション)\n\n2. **環境変数を直接使用する**:\n\n * Docker Compose コマンドを実行するときに環境変数を設定することもできます。\n\n ```bash\n OPENAI_API_KEY=your_key MODEL_NAME=gpt-4.1-mini docker compose up\n ```\n\n#### Neo4j の設定\n\nDocker Compose セットアップには、次のデフォルト構成の Neo4j コンテナーが含まれています。\n\n* ユーザー名: `neo4j`\n* パスワード: `demodemo`\n* URI: `bolt://neo4j:7687` (Docker ネットワーク内から)\n* 開発用途に最適化されたメモリ設定\n\n#### Docker Composeで実行する\n\nDocker Compose を使用してサービスを開始します。\n\n```bash\ndocker compose up\n```\n\nまたは、古いバージョンの Docker Compose を使用している場合:\n\n```bash\ndocker-compose up\n```\n\nこれにより、Neo4jデータベースとGraphiti MCPサーバーの両方が起動します。Dockerのセットアップは以下のとおりです。\n\n* パッケージ管理とサーバーの実行には`uv`を使用します\n* `pyproject.toml`ファイルから依存関係をインストールします\n* 環境変数を使用してNeo4jコンテナに接続する\n* HTTPベースのSSEトランスポート用にポート8000でサーバーを公開します\n* MCP サーバーを起動する前に、Neo4j が完全に機能していることを確認するためのヘルスチェックが含まれています。\n\n## MCPクライアントとの統合\n\n### 構成\n\nGraphiti MCP サーバーを MCP 互換クライアントで使用するには、サーバーに接続するように構成します。\n\n> \\[!重要] Pythonパッケージマネージャー`uv`がインストールされている必要があります。uv [`uv`インストール手順](https://docs.astral.sh/uv/getting-started/installation/)を参照してください。\n>\n> `uv`バイナリと Graphiti プロジェクトフォルダーへの完全なパスが設定されていることを確認してください。\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"stdio\",\n \"command\": \"/Users//.local/bin/uv\",\n \"args\": [\n \"run\",\n \"--isolated\",\n \"--directory\",\n \"/Users/>/dev/zep/graphiti/mcp_server\",\n \"--project\",\n \".\",\n \"graphiti_mcp_server.py\",\n \"--transport\",\n \"stdio\"\n ],\n \"env\": {\n \"NEO4J_URI\": \"bolt://localhost:7687\",\n \"NEO4J_USER\": \"neo4j\",\n \"NEO4J_PASSWORD\": \"password\",\n \"OPENAI_API_KEY\": \"sk-XXXXXXXX\",\n \"MODEL_NAME\": \"gpt-4.1-mini\"\n }\n }\n }\n}\n```\n\nSSE トランスポート (HTTP ベース) の場合、次の構成を使用できます。\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"sse\",\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n## 利用可能なツール\n\nGraphiti MCP サーバーは次のツールを公開します。\n\n* `add_episode` : ナレッジグラフにエピソードを追加します（テキスト、JSON、メッセージ形式をサポート）\n* `search_nodes` : 関連するノードサマリーをナレッジグラフで検索する\n* `search_facts` : ナレッジグラフで関連する事実（エンティティ間のエッジ）を検索する\n* `delete_entity_edge` : ナレッジグラフからエンティティエッジを削除する\n* `delete_episode` : ナレッジグラフからエピソードを削除する\n* `get_entity_edge` : UUIDでエンティティエッジを取得する\n* `get_episodes` : 特定のグループの最新のエピソードを取得する\n* `clear_graph` : ナレッジグラフからすべてのデータをクリアし、インデックスを再構築します。\n* `get_status` : Graphiti MCPサーバーとNeo4j接続のステータスを取得します。\n\n## JSONデータの操作\n\nGraphiti MCPサーバーは、 `add_episode`ツールで`source=\"json\"`を指定した構造化JSONデータを処理できます。これにより、構造化データからエンティティとリレーションシップを自動的に抽出できます。\n\n```\n\nadd_episode(\nname=\"Customer Profile\",\nepisode_body=\"{\\\"company\\\": {\\\"name\\\": \\\"Acme Technologies\\\"}, \\\"products\\\": [{\\\"id\\\": \\\"P001\\\", \\\"name\\\": \\\"CloudSync\\\"}, {\\\"id\\\": \\\"P002\\\", \\\"name\\\": \\\"DataMiner\\\"}]}\",\nsource=\"json\",\nsource_description=\"CRM data\"\n)\n\n```\n\n## カーソルIDEとの統合\n\nGraphiti MCP Server を Cursor IDE と統合するには、次の手順に従います。\n\n1. SSE トランスポートを使用して Graphiti MCP サーバーを実行します。\n\n```bash\npython graphiti_mcp_server.py --transport sse --use-custom-entities --group-id \n```\n\nヒント: 名前空間グラフデータに`group_id`を指定してください。group\\_id を指定しない場合、サーバーは`group_id`として「default」を使用します。\n\nまたは\n\n```bash\ndocker compose up\n```\n\n2. Cursor を Graphiti MCP サーバーに接続するように設定します。\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n3. CursorのユーザールールにGraphitiルールを追加します。詳細は[cursor\\_rules.mdを](cursor_rules.md)参照してください。\n\n4. Cursor でエージェントセッションを開始します。\n\nこの統合により、Cursor の AI アシスタントは Graphiti のナレッジグラフ機能を通じて永続的なメモリを維持できるようになります。\n\n## Claude Desktop（Docker MCP Server）との統合\n\nGraphiti MCP ServerコンテナはSSE MCPトランスポートを使用します。Claude DesktopはSSEをネイティブにサポートしていないため、 `mcp-remote`などのゲートウェイを使用する必要があります。\n\n1. **SSE トランスポートを使用して Graphiti MCP サーバーを実行します**。\n\n ```bash\n docker compose up\n ```\n\n2. **(オプション) `mcp-remote`グローバルにインストールする**: `mcp-remote`グローバルにインストールしたい場合、または`npx`によるパッケージの取得で問題が発生する場合は、グローバルにインストールできます。それ以外の場合は、次の手順で使用`npx`が自動的に処理します。\n\n ```bash\n npm install -g mcp-remote\n ```\n\n3. **Claude Desktop を構成する**: Claude Desktop 構成ファイル (通常は`claude_desktop_config.json` ) を開き、次のように`mcpServers`セクションを追加または変更します。\n\n ```json\n {\n \"mcpServers\": {\n \"graphiti-memory\": {\n // You can choose a different name if you prefer\n \"command\": \"npx\", // Or the full path to mcp-remote if npx is not in your PATH\n \"args\": [\n \"mcp-remote\",\n \"http://localhost:8000/sse\" // Ensure this matches your Graphiti server's SSE endpoint\n ]\n }\n }\n }\n ```\n\n すでに`mcpServers`エントリがある場合は、その中に新しいキーとして`graphiti-memory` (または選択した名前) を追加します。\n\n4. 変更を有効にするには、 **Claude Desktop を再起動してください**。\n\n## 要件\n\n* Python 3.10以上\n* Neo4j データベース (バージョン 5.26 以降が必要)\n* OpenAI API キー（LLM 操作および埋め込み用）\n* MCP互換クライアント\n\n## ライセンス\n\nこのプロジェクトは、親の Graphiti プロジェクトと同じライセンスの下でライセンスされます。","Japanese","ja-JP",{"_29":58,"_36":61,"_38":62},{"_31":59,"_33":59,"_34":60},"Graphiti MCP 서버","# Graphiti MCP 서버\n\nGraphiti는 동적 환경에서 작동하는 AI 에이전트를 위해 특별히 설계된 시간 인식 지식 그래프를 구축하고 쿼리하는 프레임워크입니다. 기존의 검색 증강 생성(RAG) 방식과 달리 Graphiti는 사용자 상호작용, 정형 및 비정형 엔터프라이즈 데이터, 그리고 외부 정보를 일관성 있고 쿼리 가능한 그래프로 지속적으로 통합합니다. 이 프레임워크는 전체 그래프 재계산 없이 증분형 데이터 업데이트, 효율적인 검색, 그리고 정확한 과거 쿼리를 지원하므로, 대화형 상황 인식 AI 애플리케이션 개발에 적합합니다.\n\n이는 Graphiti를 위한 실험적인 모델 컨텍스트 프로토콜(MCP) 서버 구현입니다. MCP 서버는 MCP 프로토콜을 통해 Graphiti의 주요 기능을 제공하여 AI 비서가 Graphiti의 지식 그래프 기능과 상호 작용할 수 있도록 합니다.\n\n## 특징\n\nGraphiti MCP 서버는 Graphiti의 다음과 같은 주요 고급 기능을 제공합니다.\n\n* **에피소드 관리** : 에피소드(텍스트, 메시지 또는 JSON 데이터)를 추가, 검색 및 삭제합니다.\n* **엔티티 관리** : 지식 그래프에서 엔티티 노드 및 관계를 검색하고 관리합니다.\n* **검색 기능** : 의미 검색 및 하이브리드 검색을 사용하여 사실(에지) 및 노드 요약 검색\n* **그룹 관리** : group\\_id 필터링을 통해 관련 데이터 그룹을 구성하고 관리합니다.\n* **그래프 유지 관리** : 그래프를 지우고 인덱스를 다시 작성합니다.\n\n## Claude Desktop, Cursor 및 기타 클라이언트를 위한 빠른 시작\n\n1. Graphiti GitHub 저장소를 복제하세요\n\n지엑스피1\n\n또는\n\n```bash\ngh repo clone getzep/graphiti\n```\n\n이 디렉토리의 전체 경로를 기록해 두세요.\n\n```\ncd graphiti && pwd\n```\n\n2. [Graphiti 필수 구성 요소를](#prerequisites) 설치합니다.\n\n3. Claude, Cursor 또는 다른 MCP 클라이언트가 [Graphiti를 `stdio` 전송과 함께](#integrating-with-mcp-clients) 사용하도록 구성합니다. 클라이언트 설명서에서 MCP 구성 파일의 위치를 확인하세요.\n\n## 설치\n\n### 필수 조건\n\n1. Python 3.10 이상이 설치되어 있는지 확인하세요.\n2. 실행 중인 Neo4j 데이터베이스(버전 5.26 이상 필요)\n3. LLM 작업을 위한 OpenAI API 키\n\n### 설정\n\n1. 저장소를 복제하고 mcp\\_server 디렉토리로 이동합니다.\n2. `uv` 사용하여 가상 환경을 만들고 종속성을 설치합니다.\n\n```bash\n# Install uv if you don't have it already\ncurl -LsSf https://astral.sh/uv/install.sh | sh\n\n# Create a virtual environment and install dependencies in one step\nuv sync\n```\n\n## 구성\n\n서버는 다음과 같은 환경 변수를 사용합니다.\n\n* `NEO4J_URI` : Neo4j 데이터베이스의 URI(기본값: `bolt://localhost:7687` )\n* `NEO4J_USER` : Neo4j 사용자 이름(기본값: `neo4j` )\n* `NEO4J_PASSWORD` : Neo4j 비밀번호 (기본값: `demodemo` )\n* `OPENAI_API_KEY` : OpenAI API 키(LLM 작업에 필요)\n* `OPENAI_BASE_URL` : OpenAI API의 선택적 기본 URL\n* `MODEL_NAME` : LLM 작업에 사용할 OpenAI 모델 이름입니다.\n* `SMALL_MODEL_NAME` : 소규모 LLM 작업에 사용할 OpenAI 모델 이름입니다.\n* `LLM_TEMPERATURE` : LLM 응답의 온도(0.0-2.0).\n* `AZURE_OPENAI_ENDPOINT` : 선택적 Azure OpenAI 엔드포인트 URL\n* `AZURE_OPENAI_DEPLOYMENT_NAME` : 선택적 Azure OpenAI 배포 이름\n* `AZURE_OPENAI_API_VERSION` : 선택적 Azure OpenAI API 버전\n* `AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME` : 선택적 Azure OpenAI 임베딩 배포 이름\n* `AZURE_OPENAI_EMBEDDING_API_VERSION` : 선택적 Azure OpenAI API 버전\n* `AZURE_OPENAI_USE_MANAGED_IDENTITY` : 인증을 위해 Azure 관리 ID를 선택적으로 사용합니다.\n\n프로젝트 디렉토리의 `.env` 파일에 이러한 변수를 설정할 수 있습니다.\n\n## 서버 실행\n\n`uv` 사용하여 Graphiti MCP 서버를 직접 실행하려면:\n\n```bash\nuv run graphiti_mcp_server.py\n```\n\n옵션:\n\n```bash\nuv run graphiti_mcp_server.py --model gpt-4.1-mini --transport sse\n```\n\n사용 가능한 인수:\n\n* `--model` : `MODEL_NAME` 환경 변수를 재정의합니다.\n* `--small-model` : `SMALL_MODEL_NAME` 환경 변수를 재정의합니다.\n* `--temperature` : `LLM_TEMPERATURE` 환경 변수를 재정의합니다.\n* `--transport` : 전송 방식(sse 또는 stdio, 기본값: sse)을 선택합니다.\n* `--group-id` : 그래프의 네임스페이스를 설정합니다(선택 사항). 지정하지 않으면 기본값은 \"default\"입니다.\n* `--destroy-graph` : 설정하면 시작 시 모든 Graphiti 그래프를 파괴합니다.\n* `--use-custom-entities` : 미리 정의된 ENTITY\\_TYPES를 사용하여 엔티티 추출을 활성화합니다.\n\n### 도커 배포\n\nGraphiti MCP 서버는 Docker를 사용하여 배포할 수 있습니다. Dockerfile은 패키지 관리를 위해 `uv` 사용하여 일관된 종속성 설치를 보장합니다.\n\n#### 환경 구성\n\nDocker Compose 설치를 실행하기 전에 환경 변수를 구성해야 합니다. 두 가지 옵션이 있습니다.\n\n1. **.env 파일 사용** (권장):\n\n * 제공된 `.env.example` 파일을 복사하여 `.env` 파일을 만듭니다.\n\n ```bash\n cp .env.example .env\n ```\n\n * `.env` 파일을 편집하여 OpenAI API 키 및 기타 구성 옵션을 설정합니다.\n\n ```\n # Required for LLM operations\n OPENAI_API_KEY=your_openai_api_key_here\n MODEL_NAME=gpt-4.1-mini\n # Optional: OPENAI_BASE_URL only needed for non-standard OpenAI endpoints\n # OPENAI_BASE_URL=https://api.openai.com/v1\n ```\n\n * Docker Compose 설정은 이 파일이 있는 경우 이를 사용하도록 구성됩니다(선택 사항).\n\n2. **환경 변수를 직접 사용하기** :\n\n * Docker Compose 명령을 실행할 때 환경 변수를 설정할 수도 있습니다.\n\n ```bash\n OPENAI_API_KEY=your_key MODEL_NAME=gpt-4.1-mini docker compose up\n ```\n\n#### Neo4j 구성\n\nDocker Compose 설정에는 다음과 같은 기본 구성을 갖춘 Neo4j 컨테이너가 포함됩니다.\n\n* 사용자 이름: `neo4j`\n* 비밀번호: `demodemo`\n* URI: `bolt://neo4j:7687` (Docker 네트워크 내부에서)\n* 개발용으로 최적화된 메모리 설정\n\n#### Docker Compose로 실행\n\nDocker Compose를 사용하여 서비스를 시작합니다.\n\n```bash\ndocker compose up\n```\n\n또는 이전 버전의 Docker Compose를 사용하는 경우:\n\n```bash\ndocker-compose up\n```\n\n이렇게 하면 Neo4j 데이터베이스와 Graphiti MCP 서버가 모두 시작됩니다. Docker 설정:\n\n* 패키지 관리 및 서버 실행을 위해 `uv` 사용합니다.\n* `pyproject.toml` 파일에서 종속성을 설치합니다.\n* 환경 변수를 사용하여 Neo4j 컨테이너에 연결합니다.\n* HTTP 기반 SSE 전송을 위해 포트 8000에서 서버를 노출합니다.\n* MCP 서버를 시작하기 전에 Neo4j가 완전히 작동하는지 확인하기 위한 상태 점검이 포함되어 있습니다.\n\n## MCP 클라이언트와 통합\n\n### 구성\n\nMCP 호환 클라이언트와 함께 Graphiti MCP 서버를 사용하려면 서버에 연결하도록 구성하세요.\n\n> \\[!중요] Python 패키지 관리자인 `uv` 설치되어 있어야 합니다. [`uv` 설치 지침을](https://docs.astral.sh/uv/getting-started/installation/) 참조하세요.\n>\n> `uv` 바이너리와 Graphiti 프로젝트 폴더의 전체 경로를 설정했는지 확인하세요.\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"stdio\",\n \"command\": \"/Users//.local/bin/uv\",\n \"args\": [\n \"run\",\n \"--isolated\",\n \"--directory\",\n \"/Users/>/dev/zep/graphiti/mcp_server\",\n \"--project\",\n \".\",\n \"graphiti_mcp_server.py\",\n \"--transport\",\n \"stdio\"\n ],\n \"env\": {\n \"NEO4J_URI\": \"bolt://localhost:7687\",\n \"NEO4J_USER\": \"neo4j\",\n \"NEO4J_PASSWORD\": \"password\",\n \"OPENAI_API_KEY\": \"sk-XXXXXXXX\",\n \"MODEL_NAME\": \"gpt-4.1-mini\"\n }\n }\n }\n}\n```\n\nSSE 전송(HTTP 기반)의 경우 다음 구성을 사용할 수 있습니다.\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"sse\",\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n## 사용 가능한 도구\n\nGraphiti MCP 서버는 다음 도구를 제공합니다.\n\n* `add_episode` : 지식 그래프에 에피소드 추가(텍스트, JSON, 메시지 형식 지원)\n* `search_nodes` : 지식 그래프에서 관련 노드 요약을 검색합니다.\n* `search_facts` : 지식 그래프에서 관련 사실(엔터티 간 에지)을 검색합니다.\n* `delete_entity_edge` : 지식 그래프에서 엔티티 에지를 삭제합니다.\n* `delete_episode` : 지식 그래프에서 에피소드를 삭제합니다.\n* `get_entity_edge` : UUID로 엔티티 에지를 가져옵니다.\n* `get_episodes` : 특정 그룹의 최신 에피소드를 가져옵니다.\n* `clear_graph` : 지식 그래프에서 모든 데이터를 지우고 인덱스를 다시 작성합니다.\n* `get_status` : Graphiti MCP 서버와 Neo4j 연결의 상태를 가져옵니다.\n\n## JSON 데이터 작업\n\nGraphiti MCP 서버는 `add_episode` 도구의 `source=\"json\"` 을 사용하여 구조화된 JSON 데이터를 처리할 수 있습니다. 이를 통해 구조화된 데이터에서 엔티티와 관계를 자동으로 추출할 수 있습니다.\n\n```\n\nadd_episode(\nname=\"Customer Profile\",\nepisode_body=\"{\\\"company\\\": {\\\"name\\\": \\\"Acme Technologies\\\"}, \\\"products\\\": [{\\\"id\\\": \\\"P001\\\", \\\"name\\\": \\\"CloudSync\\\"}, {\\\"id\\\": \\\"P002\\\", \\\"name\\\": \\\"DataMiner\\\"}]}\",\nsource=\"json\",\nsource_description=\"CRM data\"\n)\n\n```\n\n## 커서 IDE와 통합\n\nGraphiti MCP 서버를 Cursor IDE와 통합하려면 다음 단계를 따르세요.\n\n1. SSE 전송을 사용하여 Graphiti MCP 서버를 실행합니다.\n\n```bash\npython graphiti_mcp_server.py --transport sse --use-custom-entities --group-id \n```\n\n힌트: 그래프 데이터의 네임스페이스를 지정하려면 `group_id` 지정하세요. `group_id` 지정하지 않으면 서버는 \"default\"를 group\\_id로 사용합니다.\n\n또는\n\n```bash\ndocker compose up\n```\n\n2. Graphiti MCP 서버에 연결하도록 커서를 구성합니다.\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n3. Graphiti 규칙을 커서 사용자 규칙에 추가하세요. 자세한 내용은 [cursor\\_rules.md를](cursor_rules.md) 참조하세요.\n\n4. Cursor에서 에이전트 세션을 시작합니다.\n\n이 통합을 통해 Cursor의 AI 어시스턴트는 Graphiti의 지식 그래프 기능을 통해 지속적인 메모리를 유지할 수 있습니다.\n\n## Claude Desktop(Docker MCP 서버)과 통합\n\nGraphiti MCP Server 컨테이너는 SSE MCP 전송을 사용합니다. Claude Desktop은 기본적으로 SSE를 지원하지 않으므로 `mcp-remote` 와 같은 게이트웨이를 사용해야 합니다.\n\n1. **SSE 전송을 사용하여 Graphiti MCP 서버를 실행합니다** .\n\n ```bash\n docker compose up\n ```\n\n2. **(선택 사항) `mcp-remote` 전역으로 설치합니다** . `mcp-remote` 전역으로 설치하고 싶거나 `npx` 에서 패키지를 가져오는 데 문제가 발생하는 경우 전역으로 설치할 수 있습니다. 그렇지 않으면 다음 단계에서 사용하는 `npx` 자동으로 처리합니다.\n\n ```bash\n npm install -g mcp-remote\n ```\n\n3. **Claude Desktop 구성** : Claude Desktop 구성 파일(일반적으로 `claude_desktop_config.json` )을 열고 다음과 같이 `mcpServers` 섹션을 추가하거나 수정합니다.\n\n ```json\n {\n \"mcpServers\": {\n \"graphiti-memory\": {\n // You can choose a different name if you prefer\n \"command\": \"npx\", // Or the full path to mcp-remote if npx is not in your PATH\n \"args\": [\n \"mcp-remote\",\n \"http://localhost:8000/sse\" // Ensure this matches your Graphiti server's SSE endpoint\n ]\n }\n }\n }\n ```\n\n 이미 `mcpServers` 항목이 있는 경우 `graphiti-memory` (또는 선택한 이름)를 새 키로 추가합니다.\n\n4. 변경 사항을 적용하려면 **Claude Desktop을 다시 시작하세요** .\n\n## 요구 사항\n\n* Python 3.10 이상\n* Neo4j 데이터베이스(버전 5.26 이상 필요)\n* OpenAI API 키(LLM 작업 및 임베딩용)\n* MCP 호환 클라이언트\n\n## 특허\n\n이 프로젝트는 부모 Graphiti 프로젝트와 동일한 라이센스에 따라 라이센스가 부여되었습니다.","Korean","ko-KR",{"_29":64,"_36":67,"_38":68},{"_31":65,"_33":65,"_34":66},"Сервер MCP Graphiti","# Сервер MCP Graphiti\n\nGraphiti — это фреймворк для построения и запроса графов знаний с учетом времени, специально разработанный для агентов ИИ, работающих в динамических средах. В отличие от традиционных методов генерации дополненного поиска (RAG), Graphiti непрерывно интегрирует пользовательские взаимодействия, структурированные и неструктурированные корпоративные данные и внешнюю информацию в связный, запрашиваемый граф. Фреймворк поддерживает инкрементные обновления данных, эффективный поиск и точные исторические запросы без необходимости полного пересчета графа, что делает его подходящим для разработки интерактивных, контекстно-зависимых приложений ИИ.\n\nЭто экспериментальная реализация сервера Model Context Protocol (MCP) для Graphiti. Сервер MCP раскрывает ключевую функциональность Graphiti через протокол MCP, позволяя помощникам ИИ взаимодействовать с возможностями графа знаний Graphiti.\n\n## Функции\n\nСервер Graphiti MCP предоставляет следующие ключевые высокоуровневые функции Graphiti:\n\n* **Управление эпизодами** : добавление, извлечение и удаление эпизодов (текст, сообщения или данные JSON)\n* **Управление сущностями** : поиск и управление узлами сущностей и связями в графе знаний.\n* **Возможности поиска** : поиск фактов (ребер) и сводок узлов с использованием семантического и гибридного поиска.\n* **Управление группами** : организация и управление группами связанных данных с помощью фильтрации group\\_id.\n* **Обслуживание графика** : очистка графика и перестроение индексов.\n\n## Быстрый старт для Claude Desktop, Cursor и других клиентов\n\n1. Клонируйте репозиторий Graphiti GitHub\n\n```bash\ngit clone https://github.com/getzep/graphiti.git\n```\n\nили\n\n```bash\ngh repo clone getzep/graphiti\n```\n\nЗапомните полный путь к этому каталогу.\n\n```\ncd graphiti && pwd\n```\n\n2. Установите [необходимые компоненты Graphiti](#prerequisites) .\n\n3. Настройте Claude, Cursor или другой клиент MCP для использования [Graphiti с транспортом `stdio`](#integrating-with-mcp-clients) . См. документацию клиента о том, где найти файлы конфигурации MCP.\n\n## Установка\n\n### Предпосылки\n\n1. Убедитесь, что у вас установлен Python 3.10 или выше.\n2. Работающая база данных Neo4j (требуется версия 5.26 или более поздняя)\n3. Ключ API OpenAI для операций LLM\n\n### Настраивать\n\n1. Клонируйте репозиторий и перейдите в каталог mcp\\_server.\n2. Используйте `uv` для создания виртуальной среды и установки зависимостей:\n\n```bash\n# Install uv if you don't have it already\ncurl -LsSf https://astral.sh/uv/install.sh | sh\n\n# Create a virtual environment and install dependencies in one step\nuv sync\n```\n\n## Конфигурация\n\nСервер использует следующие переменные среды:\n\n* `NEO4J_URI` : URI для базы данных Neo4j (по умолчанию: `bolt://localhost:7687` )\n* `NEO4J_USER` : имя пользователя Neo4j (по умолчанию: `neo4j` )\n* `NEO4J_PASSWORD` : пароль Neo4j (по умолчанию: `demodemo` )\n* `OPENAI_API_KEY` : ключ API OpenAI (требуется для операций LLM)\n* `OPENAI_BASE_URL` : дополнительный базовый URL-адрес для OpenAI API.\n* `MODEL_NAME` : имя модели OpenAI, используемое для операций LLM.\n* `SMALL_MODEL_NAME` : имя модели OpenAI, используемое для небольших операций LLM.\n* `LLM_TEMPERATURE` : Температура для ответов LLM (0,0–2,0).\n* `AZURE_OPENAI_ENDPOINT` : Необязательный URL-адрес конечной точки Azure OpenAI\n* `AZURE_OPENAI_DEPLOYMENT_NAME` : Необязательное имя развертывания Azure OpenAI\n* `AZURE_OPENAI_API_VERSION` : Необязательная версия API Azure OpenAI\n* `AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME` : Необязательное имя развертывания внедрения Azure OpenAI\n* `AZURE_OPENAI_EMBEDDING_API_VERSION` : Необязательная версия API Azure OpenAI\n* `AZURE_OPENAI_USE_MANAGED_IDENTITY` : необязательное использование управляемых удостоверений Azure для аутентификации\n\nВы можете задать эти переменные в файле `.env` в каталоге проекта.\n\n## Запуск сервера\n\nЧтобы запустить сервер Graphiti MCP напрямую с помощью `uv` :\n\n```bash\nuv run graphiti_mcp_server.py\n```\n\nС опциями:\n\n```bash\nuv run graphiti_mcp_server.py --model gpt-4.1-mini --transport sse\n```\n\nДоступные аргументы:\n\n* `--model` : переопределяет переменную среды `MODEL_NAME` .\n* `--small-model` : переопределяет переменную среды `SMALL_MODEL_NAME` .\n* `--temperature` : переопределяет переменную среды `LLM_TEMPERATURE` .\n* `--transport` : Выберите метод транспорта (sse или stdio, по умолчанию: sse)\n* `--group-id` : Установить пространство имен для графика (необязательно). Если не указано, по умолчанию используется \"default\".\n* `--destroy-graph` : Если установлено, уничтожает все графы Graphiti при запуске.\n* `--use-custom-entities` : включить извлечение сущностей с использованием предопределенных ENTITY\\_TYPES\n\n### Развертывание Docker\n\nСервер Graphiti MCP можно развернуть с помощью Docker. Dockerfile использует `uv` для управления пакетами, обеспечивая последовательную установку зависимостей.\n\n#### Конфигурация среды\n\nПеред запуском настройки Docker Compose вам необходимо настроить переменные среды. У вас есть два варианта:\n\n1. **Использование файла .env** (рекомендуется):\n\n * Скопируйте предоставленный файл `.env.example` , чтобы создать файл `.env` :\n\n ```bash\n cp .env.example .env\n ```\n\n * Отредактируйте файл `.env` , чтобы задать ключ API OpenAI и другие параметры конфигурации:\n\n ```\n # Required for LLM operations\n OPENAI_API_KEY=your_openai_api_key_here\n MODEL_NAME=gpt-4.1-mini\n # Optional: OPENAI_BASE_URL only needed for non-standard OpenAI endpoints\n # OPENAI_BASE_URL=https://api.openai.com/v1\n ```\n\n * Настройка Docker Compose настроена на использование этого файла, если он существует (это необязательно)\n\n2. **Использование переменных окружения напрямую** :\n\n * Вы также можете задать переменные среды при запуске команды Docker Compose:\n\n ```bash\n OPENAI_API_KEY=your_key MODEL_NAME=gpt-4.1-mini docker compose up\n ```\n\n#### Конфигурация Neo4j\n\nНастройка Docker Compose включает контейнер Neo4j со следующей конфигурацией по умолчанию:\n\n* Имя пользователя: `neo4j`\n* Пароль: `demodemo`\n* URI: `bolt://neo4j:7687` (из сети Docker)\n* Настройки памяти оптимизированы для использования в разработке\n\n#### Работа с Docker Compose\n\nЗапустите службы с помощью Docker Compose:\n\n```bash\ndocker compose up\n```\n\nИли, если вы используете старую версию Docker Compose:\n\n```bash\ndocker-compose up\n```\n\nЭто запустит как базу данных Neo4j, так и сервер Graphiti MCP. Настройка Docker:\n\n* Использует `uv` для управления пакетами и запуска сервера\n* Устанавливает зависимости из файла `pyproject.toml`\n* Подключается к контейнеру Neo4j с использованием переменных среды\n* Предоставляет сервер на порту 8000 для HTTP-транспорта SSE\n* Включает проверку работоспособности Neo4j, чтобы убедиться в его полной работоспособности перед запуском сервера MCP.\n\n## Интеграция с клиентами MCP\n\n### Конфигурация\n\nЧтобы использовать сервер Graphiti MCP с клиентом, совместимым с MCP, настройте его для подключения к серверу:\n\n> \\[!ВАЖНО] Вам понадобится установленный менеджер пакетов Python, `uv` . Пожалуйста, ознакомьтесь с [инструкциями по установке `uv`](https://docs.astral.sh/uv/getting-started/installation/) .\n>\n> Убедитесь, что вы указали полный путь к двоичному файлу `uv` и папке вашего проекта Graphiti.\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"stdio\",\n \"command\": \"/Users//.local/bin/uv\",\n \"args\": [\n \"run\",\n \"--isolated\",\n \"--directory\",\n \"/Users/>/dev/zep/graphiti/mcp_server\",\n \"--project\",\n \".\",\n \"graphiti_mcp_server.py\",\n \"--transport\",\n \"stdio\"\n ],\n \"env\": {\n \"NEO4J_URI\": \"bolt://localhost:7687\",\n \"NEO4J_USER\": \"neo4j\",\n \"NEO4J_PASSWORD\": \"password\",\n \"OPENAI_API_KEY\": \"sk-XXXXXXXX\",\n \"MODEL_NAME\": \"gpt-4.1-mini\"\n }\n }\n }\n}\n```\n\nДля транспорта SSE (на основе HTTP) можно использовать следующую конфигурацию:\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"sse\",\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n## Доступные инструменты\n\nСервер Graphiti MCP предоставляет следующие инструменты:\n\n* `add_episode` : Добавить эпизод в граф знаний (поддерживает текст, JSON и форматы сообщений)\n* `search_nodes` : Поиск в графе знаний соответствующих сводок узлов\n* `search_facts` : Поиск в графе знаний соответствующих фактов (ребер между сущностями)\n* `delete_entity_edge` : Удалить ребро сущности из графа знаний\n* `delete_episode` : Удалить эпизод из графика знаний\n* `get_entity_edge` : Получить ребро сущности по ее UUID\n* `get_episodes` : Получить последние эпизоды для определенной группы\n* `clear_graph` : Очистить все данные из графа знаний и перестроить индексы\n* `get_status` : Получить статус сервера Graphiti MCP и соединения Neo4j\n\n## Работа с данными JSON\n\nСервер Graphiti MCP может обрабатывать структурированные данные JSON через инструмент `add_episode` с `source=\"json\"` . Это позволяет автоматически извлекать сущности и связи из структурированных данных:\n\n```\n\nadd_episode(\nname=\"Customer Profile\",\nepisode_body=\"{\\\"company\\\": {\\\"name\\\": \\\"Acme Technologies\\\"}, \\\"products\\\": [{\\\"id\\\": \\\"P001\\\", \\\"name\\\": \\\"CloudSync\\\"}, {\\\"id\\\": \\\"P002\\\", \\\"name\\\": \\\"DataMiner\\\"}]}\",\nsource=\"json\",\nsource_description=\"CRM data\"\n)\n\n```\n\n## Интеграция с Cursor IDE\n\nЧтобы интегрировать Graphiti MCP Server с Cursor IDE, выполните следующие действия:\n\n1. Запустите сервер Graphiti MCP с использованием транспорта SSE:\n\n```bash\npython graphiti_mcp_server.py --transport sse --use-custom-entities --group-id \n```\n\nПодсказка: укажите `group_id` для пространства имен графовых данных. Если вы не укажете `group_id` , сервер будет использовать \"default\" в качестве group\\_id.\n\nили\n\n```bash\ndocker compose up\n```\n\n2. Настройте курсор для подключения к серверу Graphiti MCP.\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n3. Добавьте правила Graphiti в пользовательские правила Cursor. Подробности см. в [cursor\\_rules.md](cursor_rules.md) .\n\n4. Запустите сеанс агента в Cursor.\n\nИнтеграция позволяет помощникам ИИ в Cursor поддерживать постоянную память с помощью возможностей графа знаний Graphiti.\n\n## Интеграция с Claude Desktop (Docker MCP Server)\n\nКонтейнер Graphiti MCP Server использует транспорт SSE MCP. Claude Desktop изначально не поддерживает SSE, поэтому вам нужно будет использовать шлюз, например `mcp-remote` .\n\n1. **Запустите сервер Graphiti MCP с использованием транспорта SSE** :\n\n ```bash\n docker compose up\n ```\n\n2. **(Необязательно) Установите `mcp-remote` глобально** : если вы предпочитаете, чтобы `mcp-remote` был установлен глобально, или если у вас возникли проблемы с `npx` при загрузке пакета, вы можете установить его глобально. В противном случае `npx` (используется на следующем шаге) сделает это за вас.\n\n ```bash\n npm install -g mcp-remote\n ```\n\n3. **Настройте Claude Desktop** : откройте файл конфигурации Claude Desktop (обычно `claude_desktop_config.json` ) и добавьте или измените раздел `mcpServers` следующим образом:\n\n ```json\n {\n \"mcpServers\": {\n \"graphiti-memory\": {\n // You can choose a different name if you prefer\n \"command\": \"npx\", // Or the full path to mcp-remote if npx is not in your PATH\n \"args\": [\n \"mcp-remote\",\n \"http://localhost:8000/sse\" // Ensure this matches your Graphiti server's SSE endpoint\n ]\n }\n }\n }\n ```\n\n Если у вас уже есть запись `mcpServers` , добавьте в нее `graphiti-memory` (или выбранное вами имя) в качестве нового ключа.\n\n4. **Перезапустите Claude Desktop** , чтобы изменения вступили в силу.\n\n## Требования\n\n* Python 3.10 или выше\n* База данных Neo4j (требуется версия 5.26 или более поздняя)\n* Ключ API OpenAI (для операций LLM и встраиваний)\n* MCP-совместимый клиент\n\n## Лицензия\n\nДанный проект лицензирован по той же лицензии, что и родительский проект Graphiti.","Russian","ru-RU",{"_29":70,"_36":73,"_38":74},{"_31":71,"_33":71,"_34":72},"Graphiti MCP 服务器","# Graphiti MCP 服务器\n\nGraphiti 是一个用于构建和查询时序感知知识图谱的框架，专为在动态环境中运行的 AI 代理量身定制。与传统的检索增强生成 (RAG) 方法不同，Graphiti 持续将用户交互、结构化和非结构化企业数据以及外部信息集成到一个连贯且可查询的图中。该框架支持增量数据更新、高效检索和精确的历史查询，无需完全重新计算图谱，因此非常适合开发交互式、情境感知的 AI 应用程序。\n\n这是 Graphiti 的一个实验性模型上下文协议 (MCP) 服务器实现。MCP 服务器通过 MCP 协议公开 Graphiti 的关键功能，允许 AI 助手与 Graphiti 的知识图谱功能进行交互。\n\n## 特征\n\nGraphiti MCP 服务器公开了 Graphiti 的以下关键高级功能：\n\n* **剧集管理**：添加、检索和删除剧集（文本、消息或 JSON 数据）\n* **实体管理**：搜索和管理知识图谱中的实体节点和关系\n* **搜索功能**：使用语义和混合搜索来搜索事实（边）和节点摘要\n* **群组管理**：使用 group\\_id 过滤来组织和管理相关数据组\n* **图表维护**：清除图表并重建索引\n\n## Claude Desktop、Cursor 和其他客户端的快速入门\n\n1. 克隆 Graphiti GitHub 仓库\n\n```bash\ngit clone https://github.com/getzep/graphiti.git\n```\n\n或者\n\n```bash\ngh repo clone getzep/graphiti\n```\n\n请记下该目录的完整路径。\n\n```\ncd graphiti && pwd\n```\n\n2. 安装[Graphiti 先决条件](#prerequisites)。\n\n3. 配置 Claude、Cursor 或其他 MCP 客户端，使其能够[通过`stdio`传输使用 Graphiti](#integrating-with-mcp-clients) 。请参阅客户端文档，了解其 MCP 配置文件的所在位置。\n\n## 安装\n\n### 先决条件\n\n1. 确保您已安装 Python 3.10 或更高版本。\n2. 正在运行的 Neo4j 数据库（需要 5.26 或更高版本）\n3. 用于 LLM 操作的 OpenAI API 密钥\n\n### 设置\n\n1. 克隆存储库并导航到 mcp\\_server 目录\n2. 使用`uv`创建虚拟环境并安装依赖项：\n\n```bash\n# Install uv if you don't have it already\ncurl -LsSf https://astral.sh/uv/install.sh | sh\n\n# Create a virtual environment and install dependencies in one step\nuv sync\n```\n\n## 配置\n\n服务器使用以下环境变量：\n\n* `NEO4J_URI` ：Neo4j 数据库的 URI（默认值： `bolt://localhost:7687` ）\n* `NEO4J_USER` ：Neo4j 用户名（默认值： `neo4j` ）\n* `NEO4J_PASSWORD` ：Neo4j 密码（默认值： `demodemo` ）\n* `OPENAI_API_KEY` ：OpenAI API 密钥（LLM 操作所需）\n* `OPENAI_BASE_URL` ：OpenAI API 的可选基本 URL\n* `MODEL_NAME` ：用于 LLM 操作的 OpenAI 模型名称。\n* `SMALL_MODEL_NAME` ：用于较小 LLM 操作的 OpenAI 模型名称。\n* `LLM_TEMPERATURE` ：LLM 响应的温度（0.0-2.0）。\n* `AZURE_OPENAI_ENDPOINT` ：可选的 Azure OpenAI 端点 URL\n* `AZURE_OPENAI_DEPLOYMENT_NAME` ：可选的 Azure OpenAI 部署名称\n* `AZURE_OPENAI_API_VERSION` ：可选的 Azure OpenAI API 版本\n* `AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME` ：可选的 Azure OpenAI 嵌入部署名称\n* `AZURE_OPENAI_EMBEDDING_API_VERSION` ：可选的 Azure OpenAI API 版本\n* `AZURE_OPENAI_USE_MANAGED_IDENTITY` ：可选使用 Azure 托管身份进行身份验证\n\n您可以在项目目录中的`.env`文件中设置这些变量。\n\n## 运行服务器\n\n要使用`uv`直接运行 Graphiti MCP 服务器：\n\n```bash\nuv run graphiti_mcp_server.py\n```\n\n选项包括：\n\n```bash\nuv run graphiti_mcp_server.py --model gpt-4.1-mini --transport sse\n```\n\n可用参数：\n\n* `--model` ：覆盖`MODEL_NAME`环境变量。\n* `--small-model` ：覆盖`SMALL_MODEL_NAME`环境变量。\n* `--temperature` ：覆盖`LLM_TEMPERATURE`环境变量。\n* `--transport` ：选择传输方法（sse 或 stdio，默认值：sse）\n* `--group-id` ：设置图的命名空间（可选）。若未提供，则默认为“default”。\n* `--destroy-graph` ：如果设置，则在启动时销毁所有 Graphiti 图表。\n* `--use-custom-entities` ：使用预定义的 ENTITY\\_TYPES 启用实体提取\n\n### Docker 部署\n\nGraphiti MCP 服务器可以使用 Docker 进行部署。Dockerfile 使用`uv`进行包管理，确保依赖项安装的一致性。\n\n#### 环境配置\n\n在运行 Docker Compose 设置之前，您需要配置环境变量。您有两个选择：\n\n1. **使用 .env 文件**（推荐）：\n\n * 复制提供的`.env.example`文件以创建`.env`文件：\n\n ```bash\n cp .env.example .env\n ```\n\n * 编辑`.env`文件以设置您的 OpenAI API 密钥和其他配置选项：\n\n ```\n # Required for LLM operations\n OPENAI_API_KEY=your_openai_api_key_here\n MODEL_NAME=gpt-4.1-mini\n # Optional: OPENAI_BASE_URL only needed for non-standard OpenAI endpoints\n # OPENAI_BASE_URL=https://api.openai.com/v1\n ```\n\n * 如果存在，则 Docker Compose 设置将使用此文件（可选）\n\n2. **直接使用环境变量**：\n\n * 您还可以在运行 Docker Compose 命令时设置环境变量：\n\n ```bash\n OPENAI_API_KEY=your_key MODEL_NAME=gpt-4.1-mini docker compose up\n ```\n\n#### Neo4j 配置\n\nDocker Compose 设置包含一个 Neo4j 容器，其默认配置如下：\n\n* 用户名： `neo4j`\n* 密码： `demodemo`\n* URI： `bolt://neo4j:7687` （来自 Docker 网络内部）\n* 针对开发用途优化的内存设置\n\n#### 使用 Docker Compose 运行\n\n使用 Docker Compose 启动服务：\n\n```bash\ndocker compose up\n```\n\n或者，如果您使用的是旧版本的 Docker Compose：\n\n```bash\ndocker-compose up\n```\n\n这将启动 Neo4j 数据库和 Graphiti MCP 服务器。Docker 设置如下：\n\n* 使用`uv`进行包管理和运行服务器\n* 从`pyproject.toml`文件安装依赖项\n* 使用环境变量连接到 Neo4j 容器\n* 将服务器暴露在端口 8000 上，以进行基于 HTTP 的 SSE 传输\n* 包括 Neo4j 的健康检查，以确保其在启动 MCP 服务器之前完全正常运行\n\n## 与 MCP 客户端集成\n\n### 配置\n\n要将 Graphiti MCP 服务器与 MCP 兼容客户端一起使用，请将其配置为连接到服务器：\n\n> 重要提示：你需要安装 Python 包管理器`uv` 。请参阅[`uv`安装说明](https://docs.astral.sh/uv/getting-started/installation/)。\n>\n> 确保设置`uv`二进制文件和 Graphiti 项目文件夹的完整路径。\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"stdio\",\n \"command\": \"/Users//.local/bin/uv\",\n \"args\": [\n \"run\",\n \"--isolated\",\n \"--directory\",\n \"/Users/>/dev/zep/graphiti/mcp_server\",\n \"--project\",\n \".\",\n \"graphiti_mcp_server.py\",\n \"--transport\",\n \"stdio\"\n ],\n \"env\": {\n \"NEO4J_URI\": \"bolt://localhost:7687\",\n \"NEO4J_USER\": \"neo4j\",\n \"NEO4J_PASSWORD\": \"password\",\n \"OPENAI_API_KEY\": \"sk-XXXXXXXX\",\n \"MODEL_NAME\": \"gpt-4.1-mini\"\n }\n }\n }\n}\n```\n\n对于 SSE 传输（基于 HTTP），您可以使用此配置：\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"sse\",\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n## 可用工具\n\nGraphiti MCP 服务器公开以下工具：\n\n* `add_episode` ：向知识图谱添加一集（支持文本、JSON 和消息格式）\n* `search_nodes` ：在知识图谱中搜索相关节点摘要\n* `search_facts` ：在知识图谱中搜索相关事实（实体之间的边）\n* `delete_entity_edge` ：从知识图谱中删除实体边\n* `delete_episode` ：从知识图谱中删除一集\n* `get_entity_edge` ：通过 UUID 获取实体边\n* `get_episodes` ：获取特定群组的最新剧集\n* `clear_graph` ：清除知识图谱中的所有数据并重建索引\n* `get_status` ：获取 Graphiti MCP 服务器和 Neo4j 连接的状态\n\n## 使用 JSON 数据\n\nGraphiti MCP 服务器可以通过`add_episode`工具的`source=\"json\"`参数处理结构化 JSON 数据。这允许您从结构化数据中自动提取实体和关系：\n\n```\n\nadd_episode(\nname=\"Customer Profile\",\nepisode_body=\"{\\\"company\\\": {\\\"name\\\": \\\"Acme Technologies\\\"}, \\\"products\\\": [{\\\"id\\\": \\\"P001\\\", \\\"name\\\": \\\"CloudSync\\\"}, {\\\"id\\\": \\\"P002\\\", \\\"name\\\": \\\"DataMiner\\\"}]}\",\nsource=\"json\",\nsource_description=\"CRM data\"\n)\n\n```\n\n## 与 Cursor IDE 集成\n\n要将 Graphiti MCP Server 与 Cursor IDE 集成，请按照以下步骤操作：\n\n1. 使用 SSE 传输运行 Graphiti MCP 服务器：\n\n```bash\npython graphiti_mcp_server.py --transport sse --use-custom-entities --group-id \n```\n\n提示：为命名空间图数据指定`group_id` 。如果不指定`group_id` ，服务器将使用“default”作为 group\\_id。\n\n或者\n\n```bash\ndocker compose up\n```\n\n2. 配置 Cursor 以连接到 Graphiti MCP 服务器。\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n3. 将 Graphiti 规则添加到 Cursor 的用户规则中。详情请参阅[cursor\\_rules.md](cursor_rules.md) 。\n\n4. 在 Cursor 中启动代理会话。\n\n通过集成，Cursor 中的 AI 助手能够通过 Graphiti 的知识图谱功能维护持久内存。\n\n## 与 Claude Desktop（Docker MCP 服务器）集成\n\nGraphiti MCP Server 容器使用 SSE MCP 传输。Claude Desktop 本身不支持 SSE，因此您需要使用类似`mcp-remote`的网关。\n\n1. **使用 SSE 传输运行 Graphiti MCP 服务器**：\n\n ```bash\n docker compose up\n ```\n\n2. **（可选）全局安装`mcp-remote`** ：如果您希望全局安装`mcp-remote` ，或者在使用`npx`获取包时遇到问题，则可以全局安装。否则， `npx` （下一步使用）将为您处理。\n\n ```bash\n npm install -g mcp-remote\n ```\n\n3. **配置 Claude Desktop** ：打开您的 Claude Desktop 配置文件（通常是`claude_desktop_config.json` ）并添加或修改`mcpServers`部分，如下所示：\n\n ```json\n {\n \"mcpServers\": {\n \"graphiti-memory\": {\n // You can choose a different name if you prefer\n \"command\": \"npx\", // Or the full path to mcp-remote if npx is not in your PATH\n \"args\": [\n \"mcp-remote\",\n \"http://localhost:8000/sse\" // Ensure this matches your Graphiti server's SSE endpoint\n ]\n }\n }\n }\n ```\n\n 如果您已经有`mcpServers`条目，请在其中添加`graphiti-memory` （或您选择的名称）作为新键。\n\n4. **重新启动 Claude Desktop**以使更改生效。\n\n## 要求\n\n* Python 3.10 或更高版本\n* Neo4j 数据库（需要 5.26 或更高版本）\n* OpenAI API 密钥（用于 LLM 操作和嵌入）\n* MCP 兼容客户端\n\n## 执照\n\n该项目与母公司 Graphiti 项目采用相同的许可。","Chinese","zh-CN","githubUser","locale",{"_78":79,"_13":80,"_87":32,"_88":32,"_33":32,"_16":89,"_90":91,"_131":132,"_139":-5,"_140":141,"_142":-5,"_143":144,"_226":227,"_231":-5,"_98":208,"_232":141,"_101":233,"_234":235},"addedAt","2025-04-29T20:32:56.421721Z",[81,82,83,84,85,86],"category:knowledge-and-memory","category:rag-systems","category:databases","environment:network","hosting:remote-capable","language:python","descriptionMarkdown","descriptionPlainText",6099,"integrations",[92,103,110,117,124],{"_93":94,"_31":100,"_101":102},"brand",{"_95":96,"_36":97,"_98":99},"logoSvgPath","M13.983 11.078h2.119a.186.186 0 00.186-.185V9.006a.186.186 0 00-.186-.186h-2.119a.185.185 0 00-.185.185v1.888c0 .102.083.185.185.185m-2.954-5.43h2.118a.186.186 0 00.186-.186V3.574a.186.186 0 00-.186-.185h-2.118a.185.185 0 00-.185.185v1.888c0 .102.082.185.185.185m0 2.716h2.118a.187.187 0 00.186-.186V6.29a.186.186 0 00-.186-.185h-2.118a.185.185 0 00-.185.185v1.887c0 .102.082.185.185.186m-2.93 0h2.12a.186.186 0 00.184-.186V6.29a.185.185 0 00-.185-.185H8.1a.185.185 0 00-.185.185v1.887c0 .102.083.185.185.186m-2.964 0h2.119a.186.186 0 00.185-.186V6.29a.185.185 0 00-.185-.185H5.136a.186.186 0 00-.186.185v1.887c0 .102.084.185.186.186m5.893 2.715h2.118a.186.186 0 00.186-.185V9.006a.186.186 0 00-.186-.186h-2.118a.185.185 0 00-.185.185v1.888c0 .102.082.185.185.185m-2.93 0h2.12a.185.185 0 00.184-.185V9.006a.185.185 0 00-.184-.186h-2.12a.185.185 0 00-.184.185v1.888c0 .102.083.185.185.185m-2.964 0h2.119a.185.185 0 00.185-.185V9.006a.185.185 0 00-.184-.186h-2.12a.186.186 0 00-.186.186v1.887c0 .102.084.185.186.185m-2.92 0h2.12a.185.185 0 00.184-.185V9.006a.185.185 0 00-.184-.186h-2.12a.185.185 0 00-.184.185v1.888c0 .102.082.185.185.185M23.763 9.89c-.065-.051-.672-.51-1.954-.51-.338.001-.676.03-1.01.087-.248-1.7-1.653-2.53-1.716-2.566l-.344-.199-.226.327c-.284.438-.49.922-.612 1.43-.23.97-.09 1.882.403 2.661-.595.332-1.55.413-1.744.42H.751a.751.751 0 00-.75.748 11.376 11.376 0 00.692 4.062c.545 1.428 1.355 2.48 2.41 3.124 1.18.723 3.1 1.137 5.275 1.137.983.003 1.963-.086 2.93-.266a12.248 12.248 0 003.823-1.389c.98-.567 1.86-1.288 2.61-2.136 1.252-1.418 1.998-2.997 2.553-4.4h.221c1.372 0 2.215-.549 2.68-1.009.309-.293.55-.65.707-1.046l.098-.288Z","Docker","slug","docker","Supports deployment through Docker and Docker Compose with preconfigured containers for both the MCP server and Neo4j database","uid","lqmj3spt15",{"_93":104,"_31":108,"_101":109},{"_95":105,"_36":106,"_98":107},"M24 0v24H0V0h24ZM10.933 15.89H6.84v5.52h4.198v-.93H7.955v-1.503h2.77v-.93h-2.77v-1.224h2.978v-.934Zm2.146 0h-1.084v5.52h1.035v-3.6l2.226 3.6h1.118v-5.52h-1.036v3.686l-2.259-3.687Zm5.117 0h-1.208l1.973 5.52h1.19l1.976-5.52h-1.182l-1.352 4.085-1.397-4.086ZM5.4 19.68H3.72v1.68H5.4v-1.68Z",".ENV","dotenv","Enables configuration through .env files for setting API keys, database credentials, and other environment variables","vn5q4c9l1n",{"_93":111,"_31":115,"_101":116},{"_95":112,"_36":113,"_98":114},"M12 .297c-6.63 0-12 5.373-12 12 0 5.303 3.438 9.8 8.205 11.385.6.113.82-.258.82-.577 0-.285-.01-1.04-.015-2.04-3.338.724-4.042-1.61-4.042-1.61C4.422 18.07 3.633 17.7 3.633 17.7c-1.087-.744.084-.729.084-.729 1.205.084 1.838 1.236 1.838 1.236 1.07 1.835 2.809 1.305 3.495.998.108-.776.417-1.305.76-1.605-2.665-.3-5.466-1.332-5.466-5.93 0-1.31.465-2.38 1.235-3.22-.135-.303-.54-1.523.105-3.176 0 0 1.005-.322 3.3 1.23.96-.267 1.98-.399 3-.405 1.02.006 2.04.138 3 .405 2.28-1.552 3.285-1.23 3.285-1.23.645 1.653.24 2.873.12 3.176.765.84 1.23 1.91 1.23 3.22 0 4.61-2.805 5.625-5.475 5.92.42.36.81 1.096.81 2.22 0 1.606-.015 2.896-.015 3.286 0 .315.21.69.825.57C20.565 22.092 24 17.592 24 12.297c0-6.627-5.373-12-12-12","GitHub","github","Provides installation instructions via GitHub repo cloning for accessing the Graphiti framework","omp2032tdv",{"_93":118,"_31":122,"_101":123},{"_95":119,"_36":120,"_98":121},"M9.629 13.227c-.593 0-1.139.2-1.58.533l-2.892-1.976a2.61 2.61 0 0 0 .101-.711 2.633 2.633 0 0 0-2.629-2.629A2.632 2.632 0 0 0 0 11.073a2.632 2.632 0 0 0 2.629 2.629c.593 0 1.139-.2 1.579-.533L7.1 15.145c-.063.226-.1.465-.1.711 0 .247.037.484.1.711l-2.892 1.976a2.608 2.608 0 0 0-1.579-.533A2.632 2.632 0 0 0 0 20.639a2.632 2.632 0 0 0 2.629 2.629 2.632 2.632 0 0 0 2.629-2.629c0-.247-.037-.485-.101-.711l2.892-1.976c.441.333.987.533 1.58.533a2.633 2.633 0 0 0 2.629-2.629c0-1.45-1.18-2.629-2.629-2.629ZM16.112.732c-4.72 0-7.888 2.748-7.888 8.082v3.802a3.525 3.525 0 0 1 3.071.008v-3.81c0-3.459 1.907-5.237 4.817-5.237s4.817 1.778 4.817 5.237v8.309H24V8.814C24 3.448 20.832.732 16.112.732Z","Neo4j","neo4j","Uses Neo4j as the database backend for storing and querying the knowledge graph, supporting entity relationships and temporal data management","x377l299yi",{"_93":125,"_31":129,"_101":130},{"_95":126,"_36":127,"_98":128},"M22.2819 9.8211a5.9847 5.9847 0 0 0-.5157-4.9108 6.0462 6.0462 0 0 0-6.5098-2.9A6.0651 6.0651 0 0 0 4.9807 4.1818a5.9847 5.9847 0 0 0-3.9977 2.9 6.0462 6.0462 0 0 0 .7427 7.0966 5.98 5.98 0 0 0 .511 4.9107 6.051 6.051 0 0 0 6.5146 2.9001A5.9847 5.9847 0 0 0 13.2599 24a6.0557 6.0557 0 0 0 5.7718-4.2058 5.9894 5.9894 0 0 0 3.9977-2.9001 6.0557 6.0557 0 0 0-.7475-7.0729zm-9.022 12.6081a4.4755 4.4755 0 0 1-2.8764-1.0408l.1419-.0804 4.7783-2.7582a.7948.7948 0 0 0 .3927-.6813v-6.7369l2.02 1.1686a.071.071 0 0 1 .038.052v5.5826a4.504 4.504 0 0 1-4.4945 4.4944zm-9.6607-4.1254a4.4708 4.4708 0 0 1-.5346-3.0137l.142.0852 4.783 2.7582a.7712.7712 0 0 0 .7806 0l5.8428-3.3685v2.3324a.0804.0804 0 0 1-.0332.0615L9.74 19.9502a4.4992 4.4992 0 0 1-6.1408-1.6464zM2.3408 7.8956a4.485 4.485 0 0 1 2.3655-1.9728V11.6a.7664.7664 0 0 0 .3879.6765l5.8144 3.3543-2.0201 1.1685a.0757.0757 0 0 1-.071 0l-4.8303-2.7865A4.504 4.504 0 0 1 2.3408 7.872zm16.5963 3.8558L13.1038 8.364 15.1192 7.2a.0757.0757 0 0 1 .071 0l4.8303 2.7913a4.4944 4.4944 0 0 1-.6765 8.1042v-5.6772a.79.79 0 0 0-.407-.667zm2.0107-3.0231l-.142-.0852-4.7735-2.7818a.7759.7759 0 0 0-.7854 0L9.409 9.2297V6.8974a.0662.0662 0 0 1 .0284-.0615l4.8303-2.7866a4.4992 4.4992 0 0 1 6.6802 4.66zM8.3065 12.863l-2.02-1.1638a.0804.0804 0 0 1-.038-.0567V6.0742a4.4992 4.4992 0 0 1 7.3757-3.4537l-.142.0805L8.704 5.459a.7948.7948 0 0 0-.3927.6813zm1.0976-2.3654l2.602-1.4998 2.6069 1.4998v2.9994l-2.5974 1.4997-2.6067-1.4997Z","OpenAI","openai","Leverages OpenAI's LLM capabilities for inference operations and embeddings within the knowledge graph framework","j7rf7ttxj1","namespace",{"_133":134,"_98":138},"owner",{"_135":136},"githubAccount",{"_137":138},"login","getzep","readmeMarkdown","recentUsage",0,"redditPostExternalId","repository",{"_145":146,"_192":193,"_197":198,"_217":-5,"_218":219,"_220":-5,"_221":222},"argumentsJsonSchema",{"_147":148,"_189":190,"_154":191},"properties",{"_149":150,"_156":157,"_159":160,"_162":163,"_166":167,"_169":170,"_172":173,"_175":176,"_178":179,"_181":182,"_184":185,"_186":187},"NEO4J_URI",{"_151":152,"_31":153,"_154":155},"default","bolt://localhost:7687","URI for the Neo4j database","type","string","MODEL_NAME",{"_31":158,"_154":155},"Optional model name to use for LLM inference","NEO4J_USER",{"_151":121,"_31":161,"_154":155},"Neo4j username","NEO4J_PASSWORD",{"_151":164,"_31":165,"_154":155},"demodemo","Neo4j password","OPENAI_API_KEY",{"_31":168,"_154":155},"OpenAI API key (required for LLM operations)","OPENAI_BASE_URL",{"_31":171,"_154":155},"Optional base URL for OpenAI API","AZURE_OPENAI_ENDPOINT",{"_31":174,"_154":155},"Optional Azure OpenAI endpoint URL","AZURE_OPENAI_API_VERSION",{"_31":177,"_154":155},"Optional Azure OpenAI API version","AZURE_OPENAI_DEPLOYMENT_NAME",{"_31":180,"_154":155},"Optional Azure OpenAI deployment name","AZURE_OPENAI_USE_MANAGED_IDENTITY",{"_31":183,"_154":155},"Optional use Azure Managed Identities for authentication","AZURE_OPENAI_EMBEDDING_API_VERSION",{"_31":177,"_154":155},"AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME",{"_31":188,"_154":155},"Optional Azure OpenAI embedding deployment name","required",[166],"object","githubProject",{"_34":194,"_195":196},"# Graphiti MCP Server\n\nGraphiti is a framework for building and querying temporally-aware knowledge graphs, specifically tailored for AI agents\noperating in dynamic environments. Unlike traditional retrieval-augmented generation (RAG) methods, Graphiti\ncontinuously integrates user interactions, structured and unstructured enterprise data, and external information into a\ncoherent, queryable graph. The framework supports incremental data updates, efficient retrieval, and precise historical\nqueries without requiring complete graph recomputation, making it suitable for developing interactive, context-aware AI\napplications.\n\nThis is an experimental Model Context Protocol (MCP) server implementation for Graphiti. The MCP server exposes\nGraphiti's key functionality through the MCP protocol, allowing AI assistants to interact with Graphiti's knowledge\ngraph capabilities.\n\n## Features\n\nThe Graphiti MCP server exposes the following key high-level functions of Graphiti:\n\n- **Episode Management**: Add, retrieve, and delete episodes (text, messages, or JSON data)\n- **Entity Management**: Search and manage entity nodes and relationships in the knowledge graph\n- **Search Capabilities**: Search for facts (edges) and node summaries using semantic and hybrid search\n- **Group Management**: Organize and manage groups of related data with group_id filtering\n- **Graph Maintenance**: Clear the graph and rebuild indices\n\n## Quick Start\n\n### Clone the Graphiti GitHub repo\n\n```bash\ngit clone https://github.com/getzep/graphiti.git\n```\n\nor\n\n```bash\ngh repo clone getzep/graphiti\n```\n\n### For Claude Desktop and other `stdio` only clients\n\n1. Note the full path to this directory.\n\n```\ncd graphiti && pwd\n```\n\n2. Install the [Graphiti prerequisites](#prerequisites).\n\n3. Configure Claude, Cursor, or other MCP client to use [Graphiti with a `stdio` transport](#integrating-with-mcp-clients). See the client documentation on where to find their MCP configuration files.\n\n### For Cursor and other `sse`-enabled clients\n\n1. Change directory to the `mcp_server` directory\n\n`cd graphiti/mcp_server`\n\n2. Start the service using Docker Compose\n\n`docker compose up`\n\n3. Point your MCP client to `http://localhost:8000/sse`\n\n## Installation\n\n### Prerequisites\n\n1. Ensure you have Python 3.10 or higher installed.\n2. A running Neo4j database (version 5.26 or later required)\n3. OpenAI API key for LLM operations\n\n### Setup\n\n1. Clone the repository and navigate to the mcp_server directory\n2. Use `uv` to create a virtual environment and install dependencies:\n\n```bash\n# Install uv if you don't have it already\ncurl -LsSf https://astral.sh/uv/install.sh | sh\n\n# Create a virtual environment and install dependencies in one step\nuv sync\n```\n\n## Configuration\n\nThe server uses the following environment variables:\n\n- `NEO4J_URI`: URI for the Neo4j database (default: `bolt://localhost:7687`)\n- `NEO4J_USER`: Neo4j username (default: `neo4j`)\n- `NEO4J_PASSWORD`: Neo4j password (default: `demodemo`)\n- `OPENAI_API_KEY`: OpenAI API key (required for LLM operations)\n- `OPENAI_BASE_URL`: Optional base URL for OpenAI API\n- `MODEL_NAME`: OpenAI model name to use for LLM operations.\n- `SMALL_MODEL_NAME`: OpenAI model name to use for smaller LLM operations.\n- `LLM_TEMPERATURE`: Temperature for LLM responses (0.0-2.0).\n- `AZURE_OPENAI_ENDPOINT`: Optional Azure OpenAI LLM endpoint URL\n- `AZURE_OPENAI_DEPLOYMENT_NAME`: Optional Azure OpenAI LLM deployment name\n- `AZURE_OPENAI_API_VERSION`: Optional Azure OpenAI LLM API version\n- `AZURE_OPENAI_EMBEDDING_API_KEY`: Optional Azure OpenAI Embedding deployment key (if other than `OPENAI_API_KEY`)\n- `AZURE_OPENAI_EMBEDDING_ENDPOINT`: Optional Azure OpenAI Embedding endpoint URL\n- `AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME`: Optional Azure OpenAI embedding deployment name\n- `AZURE_OPENAI_EMBEDDING_API_VERSION`: Optional Azure OpenAI API version\n- `AZURE_OPENAI_USE_MANAGED_IDENTITY`: Optional use Azure Managed Identities for authentication\n- `SEMAPHORE_LIMIT`: Episode processing concurrency. See [Concurrency and LLM Provider 429 Rate Limit Errors](#concurrency-and-llm-provider-429-rate-limit-errors)\n\nYou can set these variables in a `.env` file in the project directory.\n\n## Running the Server\n\nTo run the Graphiti MCP server directly using `uv`:\n\n```bash\nuv run graphiti_mcp_server.py\n```\n\nWith options:\n\n```bash\nuv run graphiti_mcp_server.py --model gpt-4.1-mini --transport sse\n```\n\nAvailable arguments:\n\n- `--model`: Overrides the `MODEL_NAME` environment variable.\n- `--small-model`: Overrides the `SMALL_MODEL_NAME` environment variable.\n- `--temperature`: Overrides the `LLM_TEMPERATURE` environment variable.\n- `--transport`: Choose the transport method (sse or stdio, default: sse)\n- `--group-id`: Set a namespace for the graph (optional). If not provided, defaults to \"default\".\n- `--destroy-graph`: If set, destroys all Graphiti graphs on startup.\n- `--use-custom-entities`: Enable entity extraction using the predefined ENTITY_TYPES\n\n### Concurrency and LLM Provider 429 Rate Limit Errors\n\nGraphiti's ingestion pipelines are designed for high concurrency, controlled by the `SEMAPHORE_LIMIT` environment variable.\nBy default, `SEMAPHORE_LIMIT` is set to `10` concurrent operations to help prevent `429` rate limit errors from your LLM provider. If you encounter such errors, try lowering this value.\n\nIf your LLM provider allows higher throughput, you can increase `SEMAPHORE_LIMIT` to boost episode ingestion performance.\n\n### Docker Deployment\n\nThe Graphiti MCP server can be deployed using Docker. The Dockerfile uses `uv` for package management, ensuring\nconsistent dependency installation.\n\n#### Environment Configuration\n\nBefore running the Docker Compose setup, you need to configure the environment variables. You have two options:\n\n1. **Using a .env file** (recommended):\n\n - Copy the provided `.env.example` file to create a `.env` file:\n ```bash\n cp .env.example .env\n ```\n - Edit the `.env` file to set your OpenAI API key and other configuration options:\n ```\n # Required for LLM operations\n OPENAI_API_KEY=your_openai_api_key_here\n MODEL_NAME=gpt-4.1-mini\n # Optional: OPENAI_BASE_URL only needed for non-standard OpenAI endpoints\n # OPENAI_BASE_URL=https://api.openai.com/v1\n ```\n - The Docker Compose setup is configured to use this file if it exists (it's optional)\n\n2. **Using environment variables directly**:\n - You can also set the environment variables when running the Docker Compose command:\n ```bash\n OPENAI_API_KEY=your_key MODEL_NAME=gpt-4.1-mini docker compose up\n ```\n\n#### Neo4j Configuration\n\nThe Docker Compose setup includes a Neo4j container with the following default configuration:\n\n- Username: `neo4j`\n- Password: `demodemo`\n- URI: `bolt://neo4j:7687` (from within the Docker network)\n- Memory settings optimized for development use\n\n#### Running with Docker Compose\n\nA Graphiti MCP container is available at: `zepai/knowledge-graph-mcp`. The latest build of this container is used by the Compose setup below.\n\nStart the services using Docker Compose:\n\n```bash\ndocker compose up\n```\n\nOr if you're using an older version of Docker Compose:\n\n```bash\ndocker-compose up\n```\n\nThis will start both the Neo4j database and the Graphiti MCP server. The Docker setup:\n\n- Uses `uv` for package management and running the server\n- Installs dependencies from the `pyproject.toml` file\n- Connects to the Neo4j container using the environment variables\n- Exposes the server on port 8000 for HTTP-based SSE transport\n- Includes a healthcheck for Neo4j to ensure it's fully operational before starting the MCP server\n\n## Integrating with MCP Clients\n\n### Configuration\n\nTo use the Graphiti MCP server with an MCP-compatible client, configure it to connect to the server:\n\n> [!IMPORTANT]\n> You will need the Python package manager, `uv` installed. Please refer to the [`uv` install instructions](https://docs.astral.sh/uv/getting-started/installation/).\n>\n> Ensure that you set the full path to the `uv` binary and your Graphiti project folder.\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"stdio\",\n \"command\": \"/Users//.local/bin/uv\",\n \"args\": [\n \"run\",\n \"--isolated\",\n \"--directory\",\n \"/Users/>/dev/zep/graphiti/mcp_server\",\n \"--project\",\n \".\",\n \"graphiti_mcp_server.py\",\n \"--transport\",\n \"stdio\"\n ],\n \"env\": {\n \"NEO4J_URI\": \"bolt://localhost:7687\",\n \"NEO4J_USER\": \"neo4j\",\n \"NEO4J_PASSWORD\": \"password\",\n \"OPENAI_API_KEY\": \"sk-XXXXXXXX\",\n \"MODEL_NAME\": \"gpt-4.1-mini\"\n }\n }\n }\n}\n```\n\nFor SSE transport (HTTP-based), you can use this configuration:\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"transport\": \"sse\",\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n## Available Tools\n\nThe Graphiti MCP server exposes the following tools:\n\n- `add_episode`: Add an episode to the knowledge graph (supports text, JSON, and message formats)\n- `search_nodes`: Search the knowledge graph for relevant node summaries\n- `search_facts`: Search the knowledge graph for relevant facts (edges between entities)\n- `delete_entity_edge`: Delete an entity edge from the knowledge graph\n- `delete_episode`: Delete an episode from the knowledge graph\n- `get_entity_edge`: Get an entity edge by its UUID\n- `get_episodes`: Get the most recent episodes for a specific group\n- `clear_graph`: Clear all data from the knowledge graph and rebuild indices\n- `get_status`: Get the status of the Graphiti MCP server and Neo4j connection\n\n## Working with JSON Data\n\nThe Graphiti MCP server can process structured JSON data through the `add_episode` tool with `source=\"json\"`. This\nallows you to automatically extract entities and relationships from structured data:\n\n```\n\nadd_episode(\nname=\"Customer Profile\",\nepisode_body=\"{\\\"company\\\": {\\\"name\\\": \\\"Acme Technologies\\\"}, \\\"products\\\": [{\\\"id\\\": \\\"P001\\\", \\\"name\\\": \\\"CloudSync\\\"}, {\\\"id\\\": \\\"P002\\\", \\\"name\\\": \\\"DataMiner\\\"}]}\",\nsource=\"json\",\nsource_description=\"CRM data\"\n)\n\n```\n\n## Integrating with the Cursor IDE\n\nTo integrate the Graphiti MCP Server with the Cursor IDE, follow these steps:\n\n1. Run the Graphiti MCP server using the SSE transport:\n\n```bash\npython graphiti_mcp_server.py --transport sse --use-custom-entities --group-id \n```\n\nHint: specify a `group_id` to namespace graph data. If you do not specify a `group_id`, the server will use \"default\" as the group_id.\n\nor\n\n```bash\ndocker compose up\n```\n\n2. Configure Cursor to connect to the Graphiti MCP server.\n\n```json\n{\n \"mcpServers\": {\n \"graphiti-memory\": {\n \"url\": \"http://localhost:8000/sse\"\n }\n }\n}\n```\n\n3. Add the Graphiti rules to Cursor's User Rules. See [cursor_rules.md](cursor_rules.md) for details.\n\n4. Kick off an agent session in Cursor.\n\nThe integration enables AI assistants in Cursor to maintain persistent memory through Graphiti's knowledge graph\ncapabilities.\n\n## Integrating with Claude Desktop (Docker MCP Server)\n\nThe Graphiti MCP Server container uses the SSE MCP transport. Claude Desktop does not natively support SSE, so you'll need to use a gateway like `mcp-remote`.\n\n1. **Run the Graphiti MCP server using SSE transport**:\n\n ```bash\n docker compose up\n ```\n\n2. **(Optional) Install `mcp-remote` globally**:\n If you prefer to have `mcp-remote` installed globally, or if you encounter issues with `npx` fetching the package, you can install it globally. Otherwise, `npx` (used in the next step) will handle it for you.\n\n ```bash\n npm install -g mcp-remote\n ```\n\n3. **Configure Claude Desktop**:\n Open your Claude Desktop configuration file (usually `claude_desktop_config.json`) and add or modify the `mcpServers` section as follows:\n\n ```json\n {\n \"mcpServers\": {\n \"graphiti-memory\": {\n // You can choose a different name if you prefer\n \"command\": \"npx\", // Or the full path to mcp-remote if npx is not in your PATH\n \"args\": [\n \"mcp-remote\",\n \"http://localhost:8000/sse\" // Ensure this matches your Graphiti server's SSE endpoint\n ]\n }\n }\n }\n ```\n\n If you already have an `mcpServers` entry, add `graphiti-memory` (or your chosen name) as a new key within it.\n\n4. **Restart Claude Desktop** for the changes to take effect.\n\n## Requirements\n\n- Python 3.10 or higher\n- Neo4j database (version 5.26 or later required)\n- OpenAI API key (for LLM operations and embeddings)\n- MCP-compatible client\n\n## Telemetry\n\nThe Graphiti MCP server uses the Graphiti core library, which includes anonymous telemetry collection. When you initialize the Graphiti MCP server, anonymous usage statistics are collected to help improve the framework.\n\n### What's Collected\n\n- Anonymous identifier and system information (OS, Python version)\n- Graphiti version and configuration choices (LLM provider, database backend, embedder type)\n- **No personal data, API keys, or actual graph content is ever collected**\n\n### How to Disable\n\nTo disable telemetry in the MCP server, set the environment variable:\n\n```bash\nexport GRAPHITI_TELEMETRY_ENABLED=false\n```\n\nOr add it to your `.env` file:\n\n```\nGRAPHITI_TELEMETRY_ENABLED=false\n```\n\nFor complete details about what's collected and why, see the [Telemetry section in the main Graphiti README](../README.md#telemetry).\n\n## License\n\nThis project is licensed under the same license as the parent Graphiti project.\n","url","https://github.com/getzep/graphiti/tree/main/mcp_server","githubRepository",{"_199":200,"_201":202,"_203":204,"_16":205,"_206":207,"_36":208,"_133":209,"_211":141,"_212":213,"_215":216},"defaultBranch","main","fullName","getzep/graphiti","githubTreeId",3344793,11777,"language","Python","graphiti",{"_137":138,"_101":210},"JzhbGi2wJP","recentStargazers","spdxLicense",{"_36":214},"Apache 2.0","stargazers",20351,"glamaJson","inspectable",false,"npmPackage","supportedPlatforms",[223,224,225],"WINDOWS","LINUX","MACOS","scores",{"_228":-5,"_229":-5,"_230":-5},"license","quality","security","simpleIcon","toolCount","9glqgboefx","updatedAt","2025-10-07T18:20:42.836298Z","meta",[238,241,243,250,253,255,258,261,264,267,270,272,275,278,280,282,284,286,288],{"_239":240},"title","Graphiti MCP Server | Glama",{"_242":32,"_36":31},"content",{"_244":245,"_246":247,"_248":249},"href","https://glama.ai/mcp/servers/@getzep/graphiti","rel","canonical","tagName","link",{"_242":32,"_251":252},"property","og:title",{"_242":32,"_251":254},"og:description",{"_242":256,"_251":257},"https://glama.ai/logo.png","og:logo",{"_242":259,"_251":260},"https://glama.ai/mcp/servers/@getzep/graphiti/og-image","og:image",{"_242":262,"_251":263},"630","og:image:height",{"_242":265,"_251":266},"1200","og:image:width",{"_242":268,"_251":269},"website","og:type",{"_242":245,"_251":271},"og:url",{"_242":273,"_251":274},"Glama – MCP Hosting Platform","og:site_name",{"_242":276,"_36":277},"summary_large_image","twitter:card",{"_242":273,"_36":279},"twitter:title",{"_242":32,"_36":281},"twitter:description",{"_242":259,"_36":283},"twitter:image",{"_242":262,"_36":285},"twitter:image:height",{"_242":265,"_36":287},"twitter:image:width",{"_289":290},"script:ld+json",{"_291":292,"_293":294},"@context","http://schema.org","@graph",[295],{"_291":296,"_297":298,"_299":300},"https://schema.org","@type","BreadcrumbList","itemListElement",[301,308,311,315],{"_297":302,"_303":304,"_36":305,"_306":307},"ListItem","item","https://glama.ai/","Glama","position",1,{"_297":302,"_303":309,"_36":310,"_306":22},"https://glama.ai/mcp","MCP",{"_297":302,"_303":312,"_36":313,"_306":314},"https://glama.ai/mcp/servers","Servers",3,{"_297":302,"_303":245,"_36":32,"_306":316},4,"visitorIsMaintainer","routes/_public/mcp/servers/~namespace/~slug/_pages/_explorer/blob/~commit/~/_route",{"_320":321,"_29":327,"_365":366,"_236":380,"_377":404,"_405":406},"blob",{"_242":322,"_323":324,"_325":326},"from .graphiti import Graphiti\n\n__all__ = ['Graphiti']\n","fileName","__init__.py","size",55,{"_78":79,"_13":328,"_87":32,"_88":32,"_33":32,"_16":89,"_90":329,"_131":340,"_139":-5,"_140":141,"_142":-5,"_143":343,"_226":364,"_231":-5,"_98":208,"_232":141,"_101":233,"_234":235},[81,82,83,84,85,86],[330,332,334,336,338],{"_93":331,"_31":100,"_101":102},{"_95":96,"_36":97,"_98":99},{"_93":333,"_31":108,"_101":109},{"_95":105,"_36":106,"_98":107},{"_93":335,"_31":115,"_101":116},{"_95":112,"_36":113,"_98":114},{"_93":337,"_31":122,"_101":123},{"_95":119,"_36":120,"_98":121},{"_93":339,"_31":129,"_101":130},{"_95":126,"_36":127,"_98":128},{"_133":341,"_98":138},{"_135":342},{"_137":138},{"_145":344,"_192":359,"_197":360,"_217":-5,"_218":219,"_220":-5,"_221":363},{"_147":345,"_189":358,"_154":191},{"_149":346,"_156":347,"_159":348,"_162":349,"_166":350,"_169":351,"_172":352,"_175":353,"_178":354,"_181":355,"_184":356,"_186":357},{"_151":152,"_31":153,"_154":155},{"_31":158,"_154":155},{"_151":121,"_31":161,"_154":155},{"_151":164,"_31":165,"_154":155},{"_31":168,"_154":155},{"_31":171,"_154":155},{"_31":174,"_154":155},{"_31":177,"_154":155},{"_31":180,"_154":155},{"_31":183,"_154":155},{"_31":177,"_154":155},{"_31":188,"_154":155},[166],{"_34":194,"_195":196},{"_199":200,"_201":202,"_203":204,"_16":205,"_206":207,"_36":208,"_133":361,"_211":141,"_212":362,"_215":216},{"_137":138,"_101":210},{"_36":214},[223,224,225],{"_228":-5,"_229":-5,"_230":-5},"member",{"_367":368,"_371":372,"_373":374,"_203":375,"_376":-5,"_377":324,"_378":379,"_325":326,"_154":320},"commit",{"_369":370},"sha","ae78828f9c197060df0445f735c6f299ccbf1d94","contentStoredObjectPointerId",10217043,"fullPath","graphiti_core/__init__.py",3344764,"indexable","path","safeToDisplayAsText",true,[381,383,385,387,389,390,391,393,394,395,396,397,398,399,400,401,402,403],{"_239":382},"__init__.py | Graphiti MCP Server | Glama",{"_242":384,"_36":31},"Enterprise-grade security, privacy, and features like agents, MCP, prompt templates, and more.",{"_244":386,"_246":247,"_248":249},"https://glama.ai/mcp/servers/@getzep/graphiti/blob/ae78828f9c197060df0445f735c6f299ccbf1d94/graphiti_core",{"_242":388,"_251":252},"__init__.py | Graphiti MCP Server",{"_242":384,"_251":254},{"_242":256,"_251":257},{"_242":392,"_251":260},"https://glama.ai/generated-images/og?title=__init__.py+%7C+Graphiti+MCP+Server",{"_242":262,"_251":263},{"_242":265,"_251":266},{"_242":268,"_251":269},{"_242":386,"_251":271},{"_242":273,"_251":274},{"_242":276,"_36":277},{"_242":273,"_36":279},{"_242":384,"_36":281},{"_242":392,"_36":283},{"_242":262,"_36":285},{"_242":265,"_36":287},"graphiti_core","treeMembers",[407,412,416,420,424,428,432,436,440,444,448,452,456,460,464,468,472,476,478,482,486],{"_367":408,"_373":409,"_377":410,"_154":411},{"_369":370},"graphiti_core/cross_encoder","cross_encoder","tree",{"_367":413,"_373":414,"_377":415,"_154":411},{"_369":370},"graphiti_core/driver","driver",{"_367":417,"_373":418,"_377":419,"_154":411},{"_369":370},"graphiti_core/embedder","embedder",{"_367":421,"_373":422,"_377":423,"_154":411},{"_369":370},"graphiti_core/llm_client","llm_client",{"_367":425,"_373":426,"_377":427,"_154":411},{"_369":370},"graphiti_core/migrations","migrations",{"_367":429,"_373":430,"_377":431,"_154":411},{"_369":370},"graphiti_core/models","models",{"_367":433,"_373":434,"_377":435,"_154":411},{"_369":370},"graphiti_core/prompts","prompts",{"_367":437,"_373":438,"_377":439,"_154":411},{"_369":370},"graphiti_core/search","search",{"_367":441,"_373":442,"_377":443,"_154":411},{"_369":370},"graphiti_core/telemetry","telemetry",{"_367":445,"_373":446,"_377":447,"_154":411},{"_369":370},"graphiti_core/utils","utils",{"_367":449,"_373":450,"_377":451,"_154":320},{"_369":370},"graphiti_core/decorators.py","decorators.py",{"_367":453,"_373":454,"_377":455,"_154":320},{"_369":370},"graphiti_core/edges.py","edges.py",{"_367":457,"_373":458,"_377":459,"_154":320},{"_369":370},"graphiti_core/errors.py","errors.py",{"_367":461,"_373":462,"_377":463,"_154":320},{"_369":370},"graphiti_core/graphiti.py","graphiti.py",{"_367":465,"_373":466,"_377":467,"_154":320},{"_369":370},"graphiti_core/graphiti_types.py","graphiti_types.py",{"_367":469,"_373":470,"_377":471,"_154":320},{"_369":370},"graphiti_core/graph_queries.py","graph_queries.py",{"_367":473,"_373":474,"_377":475,"_154":320},{"_369":370},"graphiti_core/helpers.py","helpers.py",{"_367":477,"_373":374,"_377":324,"_154":320},{"_369":370},{"_367":479,"_373":480,"_377":481,"_154":320},{"_369":370},"graphiti_core/nodes.py","nodes.py",{"_367":483,"_373":484,"_377":485,"_154":320},{"_369":370},"graphiti_core/py.typed","py.typed",{"_367":487,"_373":488,"_377":489,"_154":320},{"_369":370},"graphiti_core/tracer.py","tracer.py","actionData","errors"]

MCP directory API