Dataassist IO

io.github.DataAssistIO由 GitHub 帳號 DataAssistIO 發布v1.0.0更新於 Oct 24, 2026

Query your org's data in natural language — read-only MCP access to SQL, NoSQL, files & warehouses.

已驗證Streamable HTTP可網頁執行資料庫檔案與儲存資料與分析

概覽

AI 產生的概覽

託管式唯讀 MCP 伺服器，讓助理用自然語言查詢組織內的資料庫、資料倉儲與檔案。

可以做什麼
  • 探索並檢視已開放資料表的欄位、型別與整理好的中繼資料
  • 在撰寫查詢前預覽少量樣本資料列
  • 對 MySQL、PostgreSQL、BigQuery 與 Redshift 執行唯讀 SELECT 查詢
  • 對 MongoDB 與 AWS DocumentDB 執行唯讀聚合管線
  • 查詢以 Iceberg 資料表形式匯入的 CSV、Excel 與透過 SFTP 送達的檔案
適用情境
適合希望非技術同仁在 Claude、ChatGPT 等 MCP 用戶端以自然語言詢問公司資料的團隊，由管理員控管資料表層級授權並以唯讀方式存取，不必發放資料庫憑證。
執行需求
遠端 Streamable HTTP 端點，無需自行安裝。需要 DataAssist-IO 帳號與組織、至少一個已連接的資料來源，以及每位使用者透過 OAuth 2.1（動態用戶端註冊）登入授權。
安裝前請注意
這是託管的多租戶服務，查詢結果與工具呼叫都會經過 DataAssist-IO 的基礎設施；README 提到稽核紀錄可選擇擷取請求與回應內容。上傳的檔案會被匯入並以其名義存放為 Iceberg 資料表。僅能存取管理員開放給所屬團隊的資料表。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 Dataassist IO，將其新增到工作區。
  2. 連接 所需的 憑證。它們會被私密儲存，絕不會在公開頁面上展示。
  3. 為需要使用其工具的對話啟用該服務。

Web executable，透過 Streamable HTTP。 遠端服務在工作區中設定後即可從網頁執行環境執行。

交給代理程式安裝

把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式，或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南，先向你展示發布者、端點和登入方式，經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。

閱讀 https://sourceweft.com/mcp/io.github.DataAssistIO%2FDataassist-IO/install.md，並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前，先告訴我它的發布者、端點和登入方式，等我確認。不要把 API 金鑰寫進命令或我會提交的檔案；問我金鑰應從哪裡取得。

檢視代理程式所依據的安裝指南

其他 MCP 客戶端

把它新增到你客戶端的 mcpServers 設定中。 客戶端會在首次連接時完成 OAuth 登入。

{
  "mcpServers": {
    "dataassist-io": {
      "type": "http",
      "url": "https://app.dataassist.io/mcp"
    }
  }
}

README

DataAssist-IO MCP Server

Connect your databases, files, and warehouses to Claude, ChatGPT, and any MCP-compatible client — and let your team query them in plain language, safely and read-only.

[Website] [MCP Registry] [Endpoint] [ChatGPT App]

DataAssist-IO is a hosted, multi-tenant Model Context Protocol server. Your organization signs up, connects its data sources, and your users query that data through natural language in their favorite LLM client. No SQL knowledge required for end users; no data copied or exposed beyond what you explicitly grant.

This is a remote MCP server — there is nothing to install or run yourself. You connect to it over an authenticated URL.

Connection

Endpointhttps://app.dataassist.io/mcp
TransportStreamable HTTP
AuthOAuth 2.1 (Dynamic Client Registration)

You'll need a DataAssist-IO account and at least one connected data source. Sign up at dataassist.io →

Connect from Claude

In Claude (Desktop or web), add a custom connector pointing at:

https://app.dataassist.io/mcp

Claude will walk you through the OAuth sign-in. After you authorize, your org's tables appear as queryable tools.

Connect from ChatGPT

DataAssist-IO is available as a published ChatGPT app — no manual setup required:

Open the DataAssist-IO app in ChatGPT →

Launch it, sign in when prompted, and start querying your data.

Connect from any MCP client

Point any client that supports the Streamable HTTP transport with OAuth at the endpoint above. The server implements OAuth discovery (/.well-known/oauth-protected-resource), so compliant clients negotiate authentication automatically.

What you can do

Once connected, the assistant can use these read-only tools against the tables your admin has exposed to you:

ToolPurpose
list_tablesDiscover the tables available to you
describe_tableInspect a table's columns and types
get_table_metadataRead curated descriptions, tags, and per-column docs
get_sample_dataPreview a handful of rows
query_dataRun a read-only SQL SELECT (MySQL, Postgres, BigQuery, Redshift)
query_collectionRun a read-only aggregation pipeline (MongoDB, AWS DocumentDB)

Every tool is read-only by construction — queries are validated as SELECT-only and run against read-only database sessions. Nothing your assistant does can modify or delete your data.

Supported data sources

  • Files: CSV, Excel, SFTP-delivered CSV
  • Databases: MySQL, PostgreSQL, MongoDB, AWS DocumentDB
  • Warehouses: Google BigQuery, Amazon Redshift

File-based sources are imported and stored as Apache Iceberg tables on your behalf; live databases and warehouses are queried in place — your data is never copied.

Security & governance

  • Per-organization isolation — every query is scoped to your org; tenants never see each other's data.
  • OAuth-authenticated — no shared API keys; identity is established per user.
  • Team-level table grants — admins control exactly which tables each team can reach.
  • Full audit trail — every tool call is logged, with optional SOC2-grade request/response capture.
  • Read-only enforcement — validated SELECT-only access plus read-only DB sessions, defense in depth.

Getting started

  1. Create an account and your organization.
  2. Connect a data source — upload a CSV or connect a database/warehouse from the dashboard.
  3. Expose tables to your team and add curated metadata (optional but improves answer quality).
  4. Add the connector in your LLM client using the endpoint above and sign in.
  5. Ask questions about your data in plain language.

Links

About this repository

This repository is the public home for the DataAssist-IO MCP server's registry listing. DataAssist-IO is a hosted service — the server runs in DataAssist-IO's infrastructure, so there is no code to clone or deploy here. For product questions or support, visit dataassist.io.

來源：readme.md，提交 5c228e3

工具

0
該伺服器需要登入，因此 SourceWeft 無法在沒有帳號的情況下讀取其工具，也未收錄。已連接帳號所能看到的內容可能因帳號及其權限而異。

版本歷史

1
  1. v1.0.0最新Sep 16, 2026

評分

已滿足適用檢查項的 75%

該評分衡量此收錄條目的完整程度，依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價，也不表示伺服器安全或目前可用：可連通性是帶日期的檢查結果，評分也不包含安裝量或使用者評價。

  • 近 30 天內可連通必需25 分
  • 已宣告認證方式必需10 分
  • 已列出工具15 分

    請讓伺服器在無憑證時回應 tools/list，或發布可在沙箱中無憑證啟動的軟體包。

  • 描述必需15 分
  • 圖示10 分
  • 文件15 分
  • 授權條款10 分

    請在儲存庫或註冊中心條目中宣告授權條款。