Adextract MCP

io.github.anuragind003由 GitHub 帳號 anuragind003 發布v1.0.0更新於 Oct 24, 2026

Search Meta, Google Ads, LinkedIn, and TikTok ad libraries plus creative analysis via MCP.

已驗證Streamable HTTP可網頁執行資料與分析商業與電商

概覽

AI 產生的概覽

託管式 MCP 伺服器,讓助理搜尋 Meta、Google、LinkedIn 與 TikTok 廣告庫,並分析廣告創意。

可以做什麼
搜尋 Meta(Facebook 與 Instagram)廣告、粉絲頁與頁面資訊,並讀取特定廣告的歐盟/英國廣告透明度詳情。搜尋 Google Ads 透明度中心的廣告與廣告主,取得創意與廣告主詳情。搜尋 LinkedIn 與 TikTok 廣告庫,探索 TikTok 廣告主,並取得 TikTok 定向與創意詳情。用 ad_reader 分析任何公開圖片或影片創意,輸出格式、受眾、情緒與原型。使用內建提示完成熱門廣告、競品標竿、頻道策略、廣告網址分析與觀察清單等工作。
適用情境
適合競品廣告研究、跨四個廣告庫的品牌監測與創意分析,用單一介面取代分別查詢各平台。適合行銷與策略工作,讓助理取得廣告資料並轉成頻道規劃。
執行需求
遠端託管端點,不需在本機安裝。必須驗證:OAuth 登入(適用支援 OAuth 的連接器),或在 Authorization 標頭送出 Bearer API 金鑰,金鑰需在服務商帳戶頁面取得。免費額度為 50 次搜尋,另有付費方案。需要能連線至該託管端點的網路。
安裝前請注意
API 金鑰是建立時僅顯示一次的機密,請勿放入共用設定或網址。超出免費額度需付費方案。ad_reader 會把你提供的創意網址送給服務商分析。查詢與結果都會經過第三方託管服務。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 Adextract MCP,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.anuragind003%2Fad-extract/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": {
    "adextract": {
      "type": "http",
      "url": "https://mcp.adextract.co/mcp"
    }
  }
}

README

Adextract MCP Server

[MCP] [License: MIT]

[Adextract — ad intelligence for AI agents]

Adextract is an MCP (Model Context Protocol) server that gives AI assistants ad intelligence. It searches the ad libraries of Meta (Facebook and Instagram), Google Ads, LinkedIn, and TikTok, and analyzes ad creatives with ad_reader.

Hosted endpoint: https://mcp.adextract.co/mcp Free tier: 50 free searches. Get an API key at adextract.co.

⭐ If Adextract helps you, star this repo. It helps other AI developers find it.


Watch the demo

[Adextract demo — ad intelligence for AI agents]

Also on YouTube.


How it works

One MCP server, one API key, four ad libraries. Your AI assistant talks MCP — Adextract does the searching and analyzing.

[Adextract architecture — one MCP server connecting AI assistants to Meta, Google Ads, LinkedIn, and TikTok]


Quickstart

  1. Get a free API key at adextract.co (account → API keys).
  2. Add the server to Cursor with the config below.
  3. Ask your AI: "Show me Nike's active Facebook ads in the US".

There is nothing to install locally. Adextract runs as a hosted MCP server, so it works in any MCP-capable client.


Use cases

  • Competitor ad intelligence - see what your competitors are running across Meta, Google, LinkedIn, and TikTok.
  • Creative analysis - ad_reader breaks any ad creative into format, audience, emotion, and archetypes.
  • Brand monitoring - track ads for any brand or domain across four ad libraries.
  • Channel strategy - built-in MCP prompts turn ad data into channel plans.

Installation

Hosted MCP (recommended)

Connect to the Adextract hosted server:

https://mcp.adextract.co/mcp

Get your API key at adextract.co (account → API keys).

Cursor

Add to ~/.cursor/mcp.json or project .cursor/mcp.json:

json
{  "mcpServers": {    "ad-extract": {      "url": "https://mcp.adextract.co/mcp",      "headers": {        "Authorization": "Bearer YOUR_API_KEY"      }    }  }}

VS Code

Add to .vscode/mcp.json:

json
{  "servers": {    "ad-extract": {      "type": "http",      "url": "https://mcp.adextract.co/mcp",      "headers": {        "Authorization": "Bearer YOUR_API_KEY"      }    }  }}

Claude Web (OAuth)

  1. Open Claude → Settings → Connectors → Add custom connector
  2. Name: ad-extract
  3. URL: https://mcp.adextract.co/mcp
  4. Sign in and approve when prompted (no API key in the URL).

Claude Desktop (mcp-remote)

json
{  "mcpServers": {    "ad-extract": {      "command": "npx",      "args": [        "-y",        "mcp-remote",        "https://mcp.adextract.co/mcp",        "--header",        "Authorization: Bearer ***"      ]    }  }}

More clients and config paths: docs/CLIENT_SETUP.md


Available Tools

ToolPlatformDescription
search_meta_adsMetaSearch Facebook & Instagram ads
search_meta_pagesMetaDiscover pages and location IDs
get_meta_page_infoMetaPage transparency and metadata
get_meta_ad_detailsMetaEU/UK ad transparency details
search_google_adsGoogleGoogle Ads Transparency Center
search_google_ads_advertisersGoogleAdvertiser and domain search
get_google_ad_detailsGoogleCreative details
get_google_ads_advertiser_infoGoogleAdvertiser profile from Search ads
search_linkedin_adsLinkedInLinkedIn Ad Library
search_tiktok_adsTikTokTikTok Ads Library
search_tiktok_advertisersTikTokTikTok advertiser discovery
get_tiktok_ad_detailsTikTokTargeting and creative details
ad_readerAllAnalyze image/video creatives (format, audience, emotion, archetypes)

Full reference: docs/TOOLS.md


Prompts & Resources

Built-in MCP prompts for common workflows:

  • adextract_top_ads_for_brand
  • adextract_competitor_benchmark
  • adextract_channel_strategy
  • adextract_analyze_ad_url
  • adextract_schedule_watchlist

Resources include agent skill instructions and prompt-pack references (adextract://skill/main, etc.).


Authentication

MethodBest for
OAuth 2.0Claude Web and OAuth-capable connectors
Bearer API keyCursor, VS Code, Claude Desktop, mcp-remote

Discovery endpoints:

  • https://mcp.adextract.co/.well-known/oauth-protected-resource
  • https://mcp.adextract.co/.well-known/oauth-authorization-server/oauth

Server card (registry / client discovery):

  • https://mcp.adextract.co/.well-known/mcp.json

Example prompts

Show me the latest Nike ads on Facebook in the US.→ search_meta_ads({ q: "nike", country: "US", active_status: "active" })
Find Google ads for tesla.com from the last 30 days.→ search_google_ads({ domain: "tesla.com", time_period: "last_30_days" })
Analyze this ad creative URL with ad_reader.→ ad_reader({ url: "https://..." })

FAQ

What is Adextract?

Adextract is a hosted MCP server that lets AI assistants search the public ad libraries of Meta, Google Ads, LinkedIn, and TikTok, and analyze ad creatives. It works with Claude, Cursor, VS Code, and any MCP-capable client.

Which ad libraries does it cover?

Meta Ad Library (Facebook and Instagram), Google Ads Transparency Center, LinkedIn Ad Library, and TikTok Ads Library.

How do I get an API key?

Sign up at adextract.co, then go to account → API keys. The key is shown once at creation.

Is there a free tier?

Yes. New accounts get 50 free searches. Paid plans are available at adextract.co.

What does ad_reader do?

It analyzes any public image or video ad creative and returns structured JSON: format, target audience, emotion (Plutchik/VAD), archetypes, and industry/brand.

Do I need to run any code locally?

No. Adextract is a hosted server. You only add a config entry to your client.

Can I use OAuth instead of an API key?

Yes. Claude Web and other OAuth-capable connectors can sign in directly, with no API key in the URL. Other clients use a Bearer API key.

How is this different from querying the ad libraries directly?

Each ad platform has its own library, API, and access rules. Adextract gives one MCP interface across all four, plus creative analysis, without writing platform-specific code.


Documentation

DocPurpose
docs/CLIENT_SETUP.mdPer-client configuration
docs/TOOLS.mdTools, prompts, resources
docs/TROUBLESHOOTING.mdCommon errors and fixes
REGISTRY_CHECKLIST.mdPre-submission checklist
SECURITY.mdSecurity policy
SUPPORT.mdHow to get help

Links


Community

[YouTube]  [Instagram]  [LinkedIn]  [X]

Follow us for ad intelligence tips, MCP updates, and what your competitors are running.


[Adextract mascot]
Built with ❤️ by Adextract

來源:README.md,提交 ea70427

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 85%

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