PromptQL MCP Server

Name: PromptQL MCP Server
Author: hasura

@hasura

リポジトリ

PromptQL MCP Server について

Model Context Protocol (MCP) server for Hasura PromptQL

基本情報

カテゴリ

その他

ライセンス

MIT

ランタイム

python

トランスポート

stdio

公開者

hasura

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": {
    "promptql-mcp": {
      "command": "python",
      "args": [
        "-m",
        "venv",
        "venv"
      ]
    }
  }
}

ツール

ツールは検出されませんでした

ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。

概要

What is PromptQL MCP Server?

It bridges Hasura’s PromptQL data agent with AI assistants via the Model Context Protocol (MCP), enabling natural language queries against enterprise data. The server is currently unstable and only for PromptQL v1; an alpha release for v2 is coming soon.

How to use PromptQL MCP Server?

Install from source by cloning the repository, creating a Python 3.10+ virtual environment, and running pip install -e .. Configure credentials with python -m promptql_mcp_server setup (using flags for API key, playground URL, auth token, and auth mode) or via environment variables (PROMPTQL_API_KEY, PROMPTQL_PLAYGROUND_URL, PROMPTQL_AUTH_TOKEN, PROMPTQL_AUTH_MODE). For Claude Desktop, add the server to the MCP configuration JSON, pointing to the Python executable.

Key features of PromptQL MCP Server

Natural language data queries in plain English
Table artifact support for formatted results
Secure configuration for API credentials
Dual authentication modes (public and private DDN)
Data analysis and visualization insights
Simple integration with MCP-compatible clients

Use cases of PromptQL MCP Server

Ask “What were our total sales last quarter?”
Request “Who are our top five customers by revenue?”
Analyze “Show me the trend of new user signups over the past 6 months”
Query “Which products have the highest profit margin?”
Perform multi-turn conversational data analysis with context retention

FAQ from PromptQL MCP Server

What authentication modes does PromptQL MCP Server support?

It supports two modes: public (default, uses Auth-Token header for public DDN endpoints) and private (uses x-hasura-ddn-token header for private/internal DDN deployments).

What are the prerequisites for using PromptQL MCP Server?

Python 3.10 or higher, a Hasura PromptQL project with API key, playground URL, and DDN Auth Token, and an MCP-compatible client such as Claude Desktop.

How do I configure credentials for PromptQL MCP Server?

You can run python -m promptql_mcp_server setup with --api-key, --playground-url, --auth-token, and --auth-mode flags, or set the environment variables PROMPTQL_API_KEY, PROMPTQL_PLAYGROUND_URL, PROMPTQL_AUTH_TOKEN, and PROMPTQL_AUTH_MODE.

What tools does PromptQL MCP Server expose?

It exposes start_thread, start_thread_without_polling, continue_thread, get_thread_status, cancel_thread, setup_config, and check_config as MCP tools, plus a data_analysis prompt.

How do I switch between public and private authentication modes?

Reconfigure the server using the --auth-mode flag (public or private) in the setup command or update the PROMPTQL_AUTH_MODE environment variable accordingly.