VerifEye

ai.realeyes由 realeyes.ai 發布官方發布者v1.1.1更新於 Oct 24, 2026

VerifEye Face API: Identify (apparent age, gender, emotions, attention), Onboard and Compare faces.

已驗證Streamable HTTP可網頁執行安全與監控AI 與機器學習

概覽

AI 產生的概覽

Realeyes 遠端 VerifEye 伺服器，讓助理分析人臉的表觀年齡、性別、情緒與注意力，並登錄、比對與管理臉部集合。

可以做什麼
  • 分析圖片中人臉的表觀年齡、性別、當下情緒與注意力，可指定集合進行辨識
  • 將人臉登錄到集合，並把新圖片與某個已登錄人臉進行比對
  • 列出、建立、修改描述、重設與刪除集合
  • 依自訂識別碼尋找已登錄人臉、加上或移除該識別碼，以及刪除人臉
  • 取得短期有效的上傳連結以送出圖片
適用情境
適合想讓助理直接透過 VerifEye Face API 進行人臉分析或人臉比對、而不自行撰寫整合程式的情境，例如依照片估算族群屬性，或確認某位訪客是否已登錄在集合中。
執行需求
使用文件中提供的 VerifEye 遠端位址，透過 Streamable HTTP 連線，無需在本機安裝。需要 VerifEye Developer Console 帳號，以及支援 OAuth 2.1（授權碼加 PKCE）遠端伺服器的 MCP 用戶端；部分用戶端需要預先註冊的用戶端 ID。圖片須以公開的 JPEG 或 PNG 連結提供，或透過上傳連結工具上傳。除非指定美國區域，每次呼叫預設在歐盟區域執行。
安裝前請注意
會處理人臉生物辨識資料：須自行取得知情同意，並遵守 GDPR、BIPA、CCPA 等隱私與生物辨識法規。登錄會保存持久的人臉範本；刪除人臉、重設或刪除集合皆無法復原。年齡與性別僅為估算，比對結果只是相似度證據，並非身分驗證。VerifEye 會自行抓取你提供的圖片連結，建議使用短期有效的預簽章連結。呼叫以你的帳號執行並計入帳號用量上限。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 VerifEye，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/ai.realeyes%2Fverifeye/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": {
    "verifeye": {
      "type": "http",
      "url": "https://verifeye-mcp-server.realeyes.ai/mcp"
    }
  }
}

README

VerifEye MCP Server

The VerifEye MCP Server is a remotely hosted Model Context Protocol (MCP) server from Realeyes. It exposes the VerifEye Face API as tools an AI assistant can call: analyze a face for apparent age and gender plus momentary emotions and attention (Identify), onboard a face so the same person is identified later (Onboard), compare an image against one onboarded face (Compare), and manage the collections onboarded faces live in.

The server is hosted by Realeyes, so there is nothing to install or run. This repository holds its public listing metadata for MCP registries: server.json.

Full documentation: https://verifeye-docs.realeyes.ai/verifeye-apis/mcp-server/

Server URL

https://verifeye-mcp-server.realeyes.ai/mcp

Transport: Streamable HTTP. This is the server's only address. If a registry or directory lists a different one, do not connect to it, and contact us.

Requirements

Authentication

The server uses OAuth 2.1 (authorization code with PKCE). When you connect, a VerifEye sign-in and consent screen opens in your browser. You never put an API key into your MCP client: your account's API key is resolved on the server side and is never exposed to the client.

The server offers no dynamic client registration. ChatGPT, VS Code and Codex identify themselves with a Client ID Metadata Document, so there is nothing to enter. Claude, Claude Code, Cursor and the Gemini CLI use a pre-registered client ID.

Connecting a client

Step-by-step setup for Claude Code, Claude, ChatGPT, VS Code, Cursor, the Gemini CLI and Codex, including the client ID each one needs, is in the documentation: Connecting a client.

For example, in VS Code add .vscode/mcp.json to your project root:

json
{  "servers": {    "verifeye": { "type": "http", "url": "https://verifeye-mcp-server.realeyes.ai/mcp" }  }}

Start the server from the MCP: List Servers command, then sign in when the browser opens.

Using the server

Once the server is connected and you have signed in, ask in plain language and the assistant picks the tools.

Give the image as a link. The image tools take a public URL of a JPEG or PNG image, which VerifEye fetches itself. An image attached to the conversation cannot be passed to a tool, so put the image somewhere reachable and give the assistant that link. Prefer a pre-signed link that expires in minutes to a durably public URL. Where the server offers verifeye_get_upload_url and your client can make HTTPS requests of its own, the client can upload the image instead. Image requirements: documentation.

Choose the region. Every call runs in the EU (eu) unless you ask for the US (us). Collections are per region, so ask for the same region each time you work with a collection.

Example prompts

The image links are placeholders: use a link to your own image. Prompts 2 to 5 build on each other and use the collection and the face that prompt 2 creates.

#PromptWhat happens
1"Estimate the apparent age and gender of the face in https://example.com/photos/visitor.jpg."verifeye_identify_face analyzes the face and returns the estimates. No face or image is stored.
2"Create a VerifEye collection called front-desk-visitors, then onboard the face in https://example.com/photos/visitor-1042.jpg into it with the external key visitor-1042."verifeye_create_collection creates the collection, and verifeye_onboard_face stores the face template with that key. Onboarding stores biometric data: see Biometric data and consent.
3"Is the person in https://example.com/photos/today.jpg already in my front-desk-visitors collection?"verifeye_identify_face with that collection answers Existing, with the faceId and the key, when the photo shows the person onboarded in prompt 2, and NotExisting for anyone else. Nothing is onboarded.
4"Compare the face in https://example.com/photos/check-in.jpg with visitor-1042 in front-desk-visitors."verifeye_get_face_by_key finds the faceId, and verifeye_compare_face returns the similarity and VerifEye's isMatch verdict.
5"Delete visitor-1042's face from front-desk-visitors."verifeye_get_face_by_key finds the face, and verifeye_delete_face destroys its template. This cannot be undone.

Many clients ask you to approve a call before it runs, for example before a face is onboarded or deleted.

Tools

Every tool takes region (eu or us, eu by default), which selects where the image is processed and where a collection lives. Collections are per region: one created in eu is not visible from us.

ToolWhat it doesStores or destroys data
verifeye_identify_faceAnalyzes the face(s) in an image, and identifies them against a collection if you name oneNo
verifeye_onboard_faceOnboards the face(s) in an image into a collection so the person is identified laterStores a face template
verifeye_compare_faceCompares the face(s) in an image against one onboarded faceNo
verifeye_delete_faceDeletes one onboarded faceIrreversible
verifeye_list_collectionsLists the collections on the account, or checks whether one existsNo
verifeye_create_collectionCreates an empty collectionNo
verifeye_set_collection_descriptionReplaces a collection's descriptionNo
verifeye_reset_collectionDeletes every face in a collection, keeping the collectionIrreversible
verifeye_delete_collectionDeletes a collection and every face in itIrreversible
verifeye_get_face_by_keyFinds an onboarded face by your own identifierNo
verifeye_set_face_keyAttaches your own identifier to an onboarded faceNo
verifeye_delete_face_keyRemoves your identifier from an onboarded faceLoses the way of finding the face
verifeye_get_upload_urlHands back a short-lived form for uploading one imageNo

Parameters, results and error codes for each tool are in the documentation.

Biometric data and consent

These tools process facial (biometric) data. You are responsible for obtaining valid, informed consent before a call runs, and for meeting the privacy and biometric-data regulations that apply to the people in your images (for example GDPR, BIPA and CCPA).

  • verifeye_onboard_face stores a face template that persists until it is deleted.
  • verifeye_delete_face, verifeye_reset_collection and verifeye_delete_collection destroy face templates permanently.
  • Age is an apparent-age estimate, not a legal proof of age. Gender is a demographic estimate, not a self-declared identity attribute. A face match is evidence of similarity, not an identity check on its own.
  • Tool calls run as your VerifEye account and count against your account's usage limits.

Links

來源：README.md，提交 40792e5

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

2
  1. v1.1.1最新Oct 16, 2026
  2. v1.1.0Oct 11, 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。