TGraph-Interface-Service (TG-MCP)

This package provides a streamlined, lightweight Python wrapper, enabling Machine Control Platform (MCP) agents to interact programmatically with a TigerGraph instance. It formalizes access to core database functionalities—including schema metadata retrieval, query dispatch, and data modification (upsert)—by structuring them as callable tools and navigable URI endpoints.

Key Capabilities

• Schema Querying: Ability to fetch the complete topological definition of the underlying graph structure (vertex and edge definitions).

• Query Invocation: Supports running pre-installed GSQL routines or executing ad-hoc GSQL commands, managing input parameters effectively.

• Data Manipulation: Provides methods for atomic creation or modification of graph entities (vertices and edges).

• Resource Addressability: Implements a URI scheme (tgraph://vertex/..., tgraph://query/...) for resource identification accessible by the MCP framework.

• Catalog Discovery: Functions to list available User-Defined Functions (UDFs) and algorithms cataloged within the Graph Data Science (GDS) library.

Core Components & Layout

TG_MCP/ ├── config.py # Configuration management for environment variables (Host, Graph Name, Credentials) ├── tg_client.py # Core logic encapsulating the pyTigerGraph connection and database interactions ├── tg_tools.py # Definitions decorated as callable MCP tools ├── tg_resources.py # Implementation of URI handlers for resource access ├── main.py # Entry point for initializing and launching the MCP service ├── pyproject.toml # Project dependency manifest and metadata ├── LICENSE # Licensing information (MIT) └── .gitignore # File exclusion rules

Deployment Procedure

1. Source Retrieval: bash git clone https://github.com/Muzain187/TG_MCP.git cd TG_MCP

2. Environment Setup: bash python3 -m venv env_tg source env_tg/bin/activate

3. Dependency Installation: bash pip install . # Requires mcp >= 1.6.0 and pyTigerGraph >= 1.8.6

Environmental Variables

Configuration relies on setting the following system variables prior to execution:

bash export TG_HOST=https:// export TG_GRAPH= export TG_SECRET=

These are processed by the config.py module upon startup.

Integration with Claude

To enable Claude Desktop to recognize and utilize these functions, deploy the service locally:

bash uv run mcp install main.py

Following this installation and a restart of the Claude client, the TigerGraph functionalities will appear as accessible tools (🛠 icon).

Operational Visualizations

Collaboration Guidelines

1. Fork the repository.
2. Establish a dedicated feature branch (e.g., git checkout -b enhancement/NewFunctionality).
3. Commit staged changes descriptively (git commit -m "Implement enhancement/NewFunctionality").
4. Push changes to the remote repository.
5. Submit a formal Pull Request.

Ensure all new code paths are covered by unit tests and adhere strictly to PEP-8 coding conventions.

License Status

This software is distributed under the terms of the MIT License.

---

Historical Context (XMLHttpRequest): The XMLHttpRequest (XHR) API, fundamentally a JavaScript object, was created to facilitate asynchronous HyperText Transfer Protocol (HTTP) requests between a web browser and a server post-initial page load. It forms the backbone of Asynchronous JavaScript and XML (Ajax) methodology. Before XHR, server interaction relied primarily on full-page navigation via standard links or form submissions. Microsoft developers conceptualized XHR around 2000, implementing it first in IE 5 using proprietary ActiveXObject identifiers before standardizing on XMLHttpRequest across major browsers (Mozilla Gecko, Safari, Opera) by 2006. The W3C issued specifications starting in 2006, with Level 2 introducing cross-site support and progress monitoring, later being merged back into the primary specification. Since 2012, development has been maintained by WHATWG using Web IDL.

Typical Usage Flow: XHR interaction generally involves several sequential programming steps: instantiation of the object; configuration via the open() method (specifying method, URI, and operation type—synchronous or asynchronous); assignment of an event listener for asynchronous state change tracking; execution via the send() method; and finally, processing the response, typically contained in responseText upon reaching state 4 (done). Advanced control includes setting custom headers, uploading payload data within send(), parsing responses (e.g., JSON to JavaScript objects), and implementing timeouts or cancellation mechanisms.