Queryquarry

io.github.QueryQuarry由 GitHub 帳號 QueryQuarry 發布v1.0.0更新於 Nov 2, 2026

Double-blind talent marketplace: AIs search anonymous opted-in candidates; reveal on consent.

已驗證Streamable HTTP可網頁執行知識與記憶商業與電商

概覽

AI 產生的概覽

面向招募人員的遠端 MCP 伺服器：讓助理搜尋匿名且已主動同意的候選人圖譜、評估候選人，並透過雙盲握手請求聯繫。

可以做什麼
  • 依技能、地點、遠端、資歷、薪資等條件搜尋人才圖譜，回傳不含身分的媒合卡片
  • 深入評估單一候選人（技能、經歷、教育背景），仍維持匿名
  • 向候選人提出聯繫請求，並查詢請求為已送出、已接受或已婉拒
  • 對已接受的聯繫回報是否促成錄用
  • 取得指定時間後新增或更新的候選人，並儲存或檢視關注清單
  • 查看人才庫彙總統計、訂閱等級與用量上限，以及完整參考文件
適用情境
適合希望讓助理直接查詢以同意為基礎的候選人資料庫、而不去擷取個人檔案或大量寄送陌生訊息的招募與尋訪人員，且能接受候選人主動回應後才揭露身分的方式。
執行需求
需要可連上網路的遠端 streamable HTTP 端點 queryquarry.com/api/mcp。需以 QueryQuarry 招募方帳號透過 OAuth 2.0 登入並於首次使用時授權；不支援原生遠端連線的用戶端可透過 mcp-remote 橋接。免費帳號包含一定額度的候選人檢視與聯繫請求，另有付費方案。
安裝前請注意
檢視候選人詳情與聯繫請求皆按次計量，受每小時與每日上限限制，並提供付費方案，使用可能產生費用。提出聯繫會把帳號中的姓名與公司告知候選人。服務涉及候選人個人資料；身分與聯絡方式在候選人同意前維持隱藏，且此服務本身並非開源。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 Queryquarry，將其新增到工作區。
  2. 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.QueryQuarry%2Fqueryquarry/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。

{
  "mcpServers": {
    "queryquarry": {
      "type": "http",
      "url": "https://queryquarry.com/api/mcp"
    }
  }
}

README

[QueryQuarry logo — an amber gem]

QueryQuarry MCP Server

The consent-based talent graph recruiters' AIs query directly.
Search an anonymous corpus of opted-in candidates in natural language — identity is revealed only when the candidate chooses to respond.

queryquarry.com · MCP docs · Blog · [email protected]

QueryQuarry is a remote (hosted) MCP server for recruiters and sourcers. Instead of scraping or spraying InMails, your AI assistant queries a structured talent graph where every profile is explicitly opted in, candidates stay anonymous until they accept contact, and outreach happens through a double-blind escrow handshake.

Connect

Claude (claude.ai / Claude Desktop): Settings → Connectors → Add custom connector → https://queryquarry.com/api/mcp

Any MCP client (via mcp-remote):

json
{  "mcpServers": {    "queryquarry": {      "command": "npx",      "args": ["-y", "mcp-remote", "https://queryquarry.com/api/mcp"]    }  }}

On first use you'll be sent through OAuth to link your recruiter account.

Tools

ToolWhat it does
search_candidatesSearch the talent graph. Returns anonymous match cards — headline, skills, seniority, location, availability, salary range. Rich filters: skills, location, remote, seniority, employment type, salary, experience, work authorization, relocation.
get_candidateEvaluate one candidate in depth (full skills, experience, education) — still anonymous. Metered.
request_contactReach out: you identify yourself, the candidate is notified and decides. If interested, they contact you quoting a one-time code. Metered.
get_contactStatus of a contact request: sent, accepted, or declined.
get_new_candidatesProfiles new or updated since a timestamp — your standing alert.
save_candidate / get_watchlistWatchlist management.
get_corpus_statsAggregate corpus stats — counts, top skills, top locations.
check_subscriptionYour tier, limits, and usage.
get_docsThe full reference, served to your AI.

Full documentation: queryquarry.com/docs/mcp

How the double-blind flow works

  1. search_candidates → anonymous match cards, ordered by recency.
  2. get_candidate → deeper evaluation, still no name or contact info.
  3. request_contact → the candidate is notified with your identity and message.
  4. The candidate decides. If interested, they reach out to you with a one-time code. The platform never exposes a candidate's identity or contact details — consent is structural, not policy.

Pricing

Free accounts include an allowance of candidate reveals and contact requests. Paid plans: queryquarry.com/#pricing.

© 2026 QueryQuarry · This repository hosts documentation for the hosted MCP server; the service implementation is not open source.

來源：README.md，提交 cf6db36

工具

11

search_candidates

Search Candidates

Search the talent graph. Returns ANONYMOUS match cards — headline, skills (with skills_matched showing which of YOUR terms hit), seniority, location, availability, salary range, and a short snippet — but NO name, full resume, or contact. Call get_candidate (a metered reveal) for the deeper anonymous profile. Filter by skills, location, seniority, salary, experience, availability, and more; sort by recency (default) or skill_match. Always paginated (max 25 per page).

輸入參數pagesortlimitskillslocationremote_oksalary_maxsalary_minavailabilityradius_milesupdated_sinceemployment_types另有 6 個

get_candidate

Evaluate Candidate

Evaluate one candidate in depth — full skills, summary, seniority, experience (titles + what they did) and education. Stays ANONYMOUS: no name, no contact, and employer names, dates, and graduation years are withheld to protect identity. Counts toward your hourly/daily reveal limit. To actually reach someone, call request_contact.

輸入參數resume_id必填candidate_id必填

request_contact

Request Contact

Reach out to one candidate. You identify yourself (your name + company, from your account); the candidate is notified and decides. If they accept, you receive an EMAIL with their contact, their message, and a unique verification code — their identity stays private until they choose to share it. METERED (counts toward your contact limit) and requires a prior get_candidate for this resume. One active offer per candidate; respect declines. get_contact also shows the status any time.

輸入參數message必填resume_id必填candidate_id必填

get_contact

Check Contact Status

Check the status of a contact request you sent: sent (awaiting), accepted (the candidate shared their contact — see shared_contact, and expect their intro email quoting the code), or declined. Identify it by contact_id or code. Also shows any hire outcome either side reported.

輸入參數codecontact_id

report_contact_outcome

Report Hire Outcome

One-tap outcome report on an accepted contact: did it lead to a hire? Optional but appreciated — it's how the marketplace measures that consented outreach beats cold outreach. Values: hired | not_hired | in_progress.

輸入參數codeoutcome必填contact_id

get_new_candidates

New Candidates Since

Get resumes new or updated since a timestamp (your standing alert). Same filters as search_candidates. Call daily with yesterday's timestamp; no duplicates.

輸入參數since必填

save_candidate

Save to Watchlist

Save a candidate resume to your watchlist with optional notes.

輸入參數notesresume_id必填candidate_id必填

get_watchlist

View Watchlist

Get your saved candidates (watchlist).

輸入參數pagestatus

get_corpus_stats

Corpus Stats

Get aggregate stats about the talent corpus (counts, top skills/locations).

無需輸入參數。

check_subscription

Check Subscription

Check your subscription tier, status, rate limit, and usage this hour.

無需輸入參數。

get_docs

Documentation

Get the full QueryQuarry reference (how it works, all tools and filters, search tips, rate limits, privacy). Call this when you need detail beyond these tool descriptions, when guiding a new user, or before composing a complex search.

無需輸入參數。

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 100%

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
  • 已列出工具必需15 分
  • 描述必需15 分
  • 圖示10 分
  • 文件15 分
  • 授權條款10 分