Graphiti MCP Pro
English | 中文
About Graphiti
Graphiti is a framework for building and querying temporally-aware knowledge graphs, specifically tailored for AI agents operating in dynamic environments. Unlike traditional retrieval-augmented generation (RAG) methods, Graphiti continuously integrates user interactions, structured and unstructured enterprise data, and external information into a coherent, queryable graph. The framework supports incremental data updates, efficient retrieval, and precise historical queries without requiring complete graph recomputation, making it suitable for developing interactive, context-aware AI applications.
This project is an enhanced memory repository MCP service and management platform based on Graphiti. Compared to the original project's MCP service, it offers the following core advantages: enhanced core capabilities, broader AI model compatibility, and comprehensive visual management interface.
Features
Enhanced Core Capabilities
Asynchronous Parallel Processing
Adding memories is the core functionality of the MCP service. We have introduced an asynchronous parallel processing mechanism based on the original implementation. The same group ID (such as different development projects) can execute up to 5 adding memory tasks in parallel, significantly improving processing efficiency.
Task Management Tools
Four new MCP tools have been added for managing add_memory tasks:
list_add_memory_tasks- List alladd_memorytasksget_add_memory_task_status- Getadd_memorytask statuswait_for_add_memory_task- Wait foradd_memorytask completioncancel_add_memory_task- Canceladd_memorytask
Unified Configuration Management
Optimized configuration management to resolve inconsistencies between command-line parameters, environment variables, and management backend database configurations.
Broader AI Model Compatibility and Flexibility
Enhanced Model Compatibility
Through integration with the instructor library, model compatibility has been significantly improved. Now supports various models such as DeepSeek, Qwen, and even locally run models through Ollama, vLLM, as long as they provide OpenAI API compatible interfaces.
Separated Model Configuration
The original unified LLM configuration has been split into three independent configurations, allowing flexible combinations based on actual needs:
- Large Model (LLM): Responsible for entity and relationship extraction
- Small Model (Small LLM): Handles entity attribute summarization, relationship deduplication, reranking, and other lightweight tasks
- Embedding Model (Embedder): Dedicated to text vectorization
Comprehensive Management Platform
To provide better user experience and observability, we have developed a complete management backend and Web UI. Through the management interface, you can:
- Service Control: Start, stop, restart MCP service
- Configuration Management: Real-time configuration updates and adjustments
- Usage Monitoring: View detailed token usage statistics
- Log Viewing: Real-time and historical log queries
Getting Started
Running with Docker Compose (Recommended)
-
Clone Project
git clone http://github.com/itcook/graphiti-mcp-pro # or git clone git@github.com:itcook/graphiti-mcp-pro.git cd graphiti-mcp-pro
-
Configure Environment Variables (Optional)
# Copy example configuration file mv .env.example.en .env # Edit .env file according to the instructions
-
Start Services
docker compose up -d
- Access Management Interface Default address: http://localhost:6062
Manual Installation
-
Clone Project
git clone http://github.com/itcook/graphiti-mcp-pro # or git clone git@github.com:itcook/graphiti-mcp-pro.git cd graphiti-mcp-pro
-
Install Dependencies
uv sync -
Configure Environment Variables
# Copy example configuration file mv .env.example.en .env # Edit .env file according to the instructions
-
Run MCP Service
# Run service with management backend uv run main.py -m # Or run MCP service only # uv run main.py
-
Build and Run Management Frontend
Enter frontend directory and install dependencies:
cd manager/frontend pnpm install # or npm install / yarn
Build and run frontend:
pnpm run build # or npm run build / yarn build pnpm run preview # or npm run preview / yarn preview
Access management interface: http://localhost:6062
Important Notes
Known Limitations
- 🔒 Security Notice: The management backend does not implement authorization access mechanisms. DO NOT expose the service on public servers.
- 🧪 Test Coverage: Due to resource constraints, the project has not been thoroughly tested. Recommended for personal use only.
- 📡 Transport Protocol: Only supports streamable-http transport protocol. Removed stdio and sse support from the original project.
- ⚙️ Code Optimization: Some architectural designs (dependency injection, exception handling, client decoupling, etc.) still have room for optimization.
Usage Recommendations
- Configuration Instructions: Please carefully read the setup instructions and comments in
.env.example.en - Model Selection: If using natively supported models like GPT/Gemini/Claude and don't need detailed runtime information, consider using the original Graphiti MCP
- Issue Feedback: Welcome to submit Issues or Pull Requests for any usage problems
Developed with assistance from 🤖 Augment Code
Metadata
Release files for iflow-mcp_itcook-mcp-server 0.4.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| iflow_mcp_itcook_mcp_server-0.4.1.tar.gz | 313.2 kB | Without provenance | Details for iflow_mcp_itcook_mcp_server-0.4.1.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| iflow_mcp_itcook_mcp_server-0.4.1-py3-none-any.whl | Python 3 | none | any | Details for iflow_mcp_itcook_mcp_server-0.4.1-py3-none-any.whl |
Total release size: 540.4 kB
Release files / iflow_mcp_itcook_mcp_server-0.4.1.tar.gz
| Download URL | iflow_mcp_itcook_mcp_server-0.4.1.tar.gz |
|---|---|
| Size | 313.2 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
e81551494626de13e7a9259dd9c49a56f1fd31b0fc01d434c728501647fc4c49
|
|
BLAKE2b-256 checksum How to use checksums |
a1b51512f0275046434cc20e3f6f69ee8e53e3c431b3ec18ca204a7c49f3f744
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
uv/0.10.0 {"installer":{"name":"uv","version":"0.10.0","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Debian GNU/Linux","version":"13","id":"trixie","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}
|
Release files / iflow_mcp_itcook_mcp_server-0.4.1-py3-none-any.whl
| Download URL | iflow_mcp_itcook_mcp_server-0.4.1-py3-none-any.whl |
|---|---|
| Size | 227.2 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
44269480830942587400c0d27678e1235deaab59550af3708a7a4b654acdf1e9
|
|
BLAKE2b-256 checksum How to use checksums |
34f6679f2670afd9350116c89ec81084762686b74aad7061ffbeb6b4f7c91449
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
uv/0.10.0 {"installer":{"name":"uv","version":"0.10.0","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Debian GNU/Linux","version":"13","id":"trixie","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}
|
Security for iflow-mcp_itcook-mcp-server 0.4.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.