Gxtract MCP Server

@sascharo

リポジトリホームページ

基本情報

カテゴリ

その他

ライセンス

NOASSERTION

トランスポート

stdio

作者

sascharo

投稿者

Sascha Robitzki

Gxtract MCP Server について

GXtract MCP Server for VS Code, forks and other editors; provides tools for interacting with GroundX

#groundx#mcp#document-understanding

Gxtract MCP Server の設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "gxtract": {
      "command": "uv",
      "type": "stdio",
      "args": [
        "--directory",
        "DRIVE:\\path\\to\\your\\gxtract",
        "--project",
        "DRIVE:\\path\\to\\your\\gxtract",
        "run",
        "gxtract",
        "--transport",
        "stdio"
      ],
      "env": {
        "GROUNDX_API_KEY": "${env:GROUNDX_API_KEY}"
      }
    }
  }
}

Gxtract MCP Server のツール

Search for documents within your GroundX projects.

Ask specific questions about a document in GroundX.

Get explanations for diagrams, tables, or other semantic objects within documents.

Manually refresh the GroundX metadata cache.

Manually refresh the GroundX projects and buckets cache.

Get statistics about the cached metadata.

List all currently cached GroundX resources (projects, buckets).

概要

What is GXtract?

GXtract is a Model Context Protocol (MCP) server that integrates with VS Code and other compatible editors, providing a suite of tools for interacting with the GroundX platform to leverage its document understanding capabilities directly within a development environment.

How to use GXtract?

Install Python 3.12+, UV 0.7.6+, and clone the repository. Run uv sync to install dependencies. Configure your GroundX API key as an environment variable or via VS Code's secure inputs, then add a server entry in VS Code's settings.json under mcp.servers with the appropriate command and args. Reload VS Code to access the tools through MCP features.

Key features of GXtract

GroundX integration for document search, querying, and semantic object explanation.
MCP compliant for use with VS Code’s MCP client and compatible systems.
Built with Python 3.12+ and FastMCP v2 for performance.
Simple setup via VS Code settings configuration.
In-memory cache for GroundX metadata to reduce API calls.

Use cases of GXtract

Searching documents within your GroundX projects.
Asking specific questions about a document in GroundX.
Getting explanations for diagrams, tables, or other semantic objects within documents.
Manually refreshing the GroundX metadata cache when needed.

よくある質問

What are the prerequisites for using GXtract?

You need Python 3.12 or higher, UV version 0.7.6 or higher, and a valid GroundX API key from the GroundX Dashboard.

How do I configure GXtract in VS Code?

Clone the repository, run `uv sync`, set your `GROUNDX_API_KEY` environment variable, then add a server entry in VS Code’s `settings.json` under `mcp.servers` with `command: "uv"`, appropriate `args` pointing to the project directory, and the transport set to `stdio`. Reload VS Code after saving.

How can I securely provide my GroundX API key?

You can set it as a system environment variable (`GROUNDX_API_KEY`) and reference it in VS Code settings with `${env:GROUNDX_API_KEY}`, or use VS Code’s secure inputs feature to prompt and store the key in your system’s credential manager. Avoid storing the key in plaintext in `settings.json`.

What transport types does GXtract support?

GXtract supports both `stdio` and `http` transport. VS Code currently only supports `stdio`. The transport can be set via the `--transport` argument or in the `args` array in VS Code settings.

How do I refresh the metadata cache?

Use the tool `cache/refreshMetadataCache` or `cache/refreshCachedResources` provided by the server. You can also disable caching entirely with the `--disable-cache` flag or adjust the cache TTL using `--cache-ttl`.

「その他」の他のコンテンツ

DeepMark

DeepMark helps teachers deliver rapid, consistent marking with meaningful feedback for every student — in a fraction of the time. What once took a week, now takes one free period.