
AdsCove
com.adscove由 adscove.com 發布官方發布者v1.0.0更新於 Oct 28, 2026
Remote MCP for Shopify, Meta Ads, Google Ads and TikTok Ads with human approval for writes
概覽
遠端 MCP 伺服器,讓助理檢視並修改 Shopify、Meta Ads、Google Ads 與 TikTok Ads 帳戶,寫入操作需人工核准。
- 可以做什麼
- 產出業務與廣告報表:Shopify 與廣告帳戶的每日報告、跨平台廣告活動指標、歸因落差、浪費花費、搜尋字詞、異常與素材疲勞分析
- 檢視並上傳廣告素材與資源,替換 Shopify 商品圖片
- 在 Meta、Google 與 TikTok 建立暫停或停用的廣告活動、廣告組合與廣告,並調整狀態、預算、出價、定向與否定關鍵字
- 管理 Meta 與 TikTok 原生自動規則,編輯 Shopify 商品、頁面、集合、折扣與庫存
- 透過精簡的漸進式工具目錄探索平台工具名稱並呼叫,待執行操作依 execution_id 核准或拒絕
- 適用情境
- 適合電商與付費媒體團隊,讓助理回答帳戶表現如何、花費浪費在哪裡、廣告為何被拒等問題,並準備廣告活動或 Shopify 變更交由人工核准。
- 執行需求
- 遠端 Streamable HTTP 端點 adscove.com/api/mcp,無需本機安裝。認證方式為 OAuth,或透過 Authorization Bearer 標頭傳入 API 金鑰。需先在 AdsCove 網頁端連接各平台帳戶,並受 AdsCove 帳戶的定價與綁定數量限制約束。
安裝
在 SourceWeft 中
- 開啟 儀表板中的 AdsCove,將其新增到工作區。
- 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。
- 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/com.adscove%2Fadscove/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。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": {
"adscove": {
"type": "http",
"url": "https://adscove.com/api/mcp"
}
}
}README
AdsCove MCP Server
AdsCove is an approval-gated advertising-operations server that connects AI agents to Shopify, Meta Ads, Google Ads, and TikTok Ads.
Endpoint
- URL:
https://adscove.com/api/mcp - Transport: Streamable HTTP. Clients should send
Accept: application/json, text/event-stream. JSON-only clients are also supported. - Authentication:
- OAuth, for clients that support remote-MCP OAuth.
- API key, for other clients:
Authorization: Bearer sk_live_.... Generate a key at https://adscove.com/keys.
- Catalog modes:
- Default (progressive core):
tools/liststays small (connections, reporting, capabilities,discover_adscove_tools, approvals). Calldiscover_adscove_toolsto find platform tool names, thenrun_adscove_tool. https://adscove.com/api/mcp?tools=expanded: the previous fat compact list.https://adscove.com/api/mcp?tools=full: every schema.
- Default (progressive core):
Setup: connect your platforms at https://adscove.com/connections, then add the endpoint as a remote MCP server in your AI client.
What it does
Platforms:
- Shopify (optional)
- Meta Ads
- Google Ads
- TikTok Ads
The launch-available registry has 329 tools (182 read, 147 write), as of the 2026-10-20 source review.
Capabilities:
- Reporting and diagnosis
- Daily business report across Shopify and ad accounts.
- Cross-platform campaign performance in one schema.
- Attribution-gap analysis, wasted-spend and search-term analysis, anomaly explanations, creative fatigue, and tracking-URL audits.
- One-call business insights for "how are my accounts doing" questions.
- Creatives
- Upload images and videos to the Meta, Google, and TikTok ad libraries.
- Inspect existing ad creatives and assets.
- Replace Shopify product images.
- Launch and optimize (writes, approval-gated)
- Create paused Meta campaigns, ad sets, and ads, and paused Google Search, Performance Max, Demand Gen, and Display drafts.
- Create disabled TikTok campaigns and ad groups.
- Change status, budgets, bids, targeting, and negative keywords.
- Create and manage native Meta and TikTok automated rules.
- Edit Shopify products, pages, collections, discounts, and inventory.
Blended ROAS in AdsCove is Shopify GMV divided by combined Meta, Google, and TikTok spend. Because it uses total store GMV, it is closer to MER than to campaign-level attribution ROAS. See https://adscove.com/guides/shopify-meta-google-roas for the formulas and limits.
Safety model
- Every real write needs explicit approval. A write returns a pending action with an
execution_id. The agent runs it withapprove_actiononly after the user consents in the conversation, or drops it withreject_action. - New ads start paused. Meta and Google creation tools create PAUSED resources. TikTok creation tools use DISABLE. Activating a resource is a separate approved action.
- Dry runs come first where they apply. Shopify media and file uploads default to a dry run.
google_launch_campaignlists exactly what it will enable before approval. Automation rules are created disabled, pass static checks and a Dry Run, and are activated separately. - Errors include
error.code,category,retryable,user_action, andoperator_action. Clients should branch onerror.codefirst. - Unconnected platforms return
not_connected. Tools without a live adapter returncapability_not_implemented, never a sample success. - Hiding an account from lists does not pause ads or disable rules or schedules.
- The security documentation covers how reads are separated from writes, how platform tokens are encrypted, how duplicate writes are prevented, and how audit history is recorded: https://adscove.com/docs/security
- AdsCove is not an autonomous agent. It does not change ads or spend budget without approval.
- AdsCove is in beta and does not provide an uptime SLA. Keep the native platform safeguards enabled.
Quick start
Connect your platforms at https://adscove.com/connections first. Then add the endpoint in your client. To check the connection, ask the agent to call list_connections.
Claude (web, desktop)
- Open Settings, then Connectors, then Add custom connector.
- Paste
https://adscove.com/api/mcpas the URL. - Sign in with OAuth when prompted. Setup guide: https://adscove.com/clients/claude
Claude Code
Setup guide: https://adscove.com/clients/claude-code
Cursor
Add this to .cursor/mcp.json:
Setup guide: https://adscove.com/clients/cursor
Codex
Add this to ~/.codex/config.toml:
Setup guide: https://adscove.com/clients/codex
ChatGPT
Custom MCP apps in ChatGPT are currently web-only. Add AdsCove as a custom MCP app on the ChatGPT web app, using https://adscove.com/api/mcp. Setup guide: https://adscove.com/clients/chatgpt
Gemini CLI
Add this to your Gemini CLI settings.json:
Setup guide: https://adscove.com/clients/gemini
The examples/ folder contains these snippets as separate files.
Example prompts
- "Compare Meta-reported ROAS with Shopify revenue for the last 7 days."
- "How are my accounts doing, and what should I optimize first?"
- "Which Google search terms are wasting spend? Suggest negative keywords, but don't apply them yet."
- "Check whether any Meta ads are disapproved or pending review, and tell me why."
- "Prepare a paused Meta test campaign for this product and show me the pending changes. Wait for my approval before running them."
Links
- Website: https://adscove.com
- Product introduction: https://adscove.com/docs
- Quickstart: https://adscove.com/docs/quickstart
- Security and approval model: https://adscove.com/docs/security
- Pricing and account binding limits: https://adscove.com/docs/pricing
- Live tool reference: https://adscove.com/docs/tools
- REST API: https://adscove.com/docs/api
- Full page index: https://adscove.com/llms-full.txt
- Guides: https://adscove.com/guides
- Meta Ads MCP setup: https://adscove.com/guides/meta-ads-mcp-setup
- Google Ads MCP setup: https://adscove.com/guides/google-ads-mcp-setup
- Shopify, Meta, and Google ROAS: https://adscove.com/guides/shopify-meta-google-roas
- Meta ROAS vs Shopify: https://adscove.com/guides/meta-roas-vs-shopify
- Approval-gated ad changes: https://adscove.com/guides/approval-gated-ad-changes
- AI ad operations safety: https://adscove.com/guides/ai-ad-operations-safety
- Claude Desktop setup: https://adscove.com/guides/claude-desktop-ecommerce-mcp
- Cursor and Claude install: https://adscove.com/guides/cursor-claude-ecommerce-mcp-install
- Pricing: https://adscove.com/pricing
License
The server is a hosted service operated by AdsCove. This repository contains documentation and configuration examples only (MIT).
來源:README.md,提交 bdedeb5
工具
0版本歷史
1- v1.0.0最新Oct 23, 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