AdSpends MCP Server|Model Context ProtocolNEW

Meta Ads MCP Server: The Facebook Ad Library inside Claude, ChatGPT & Cursor

Ask your AI for winning ads, competitor ad spend, advertorials and sales pages. It answers with live AdSpends data instead of guesses.

Works with
  • Claude
  • ChatGPT
  • Perplexity
  • Cursor
  • GitHub Copilot
  • Gemini CLI
  • Windsurf
  • 22 read-only tools: ads, brands, landing pages and your saved boards
  • Estimated EUR ad spend for every Meta ad delivered in the EU
  • Connect in under a minute: OAuth for Claude, ChatGPT and Perplexity
MCP tools
22
Brands
848,991
Ads with ad spend
8,958,549
Est. ad spend
€14.1B

Setup

Connect AdSpends to your AI in 3 steps

No installs, no Meta developer account. The whole setup takes under a minute.

  1. 1under a minute

    Connect

    Paste https://app.adspends.com/api/mcp as a custom connector in Claude, ChatGPT or Perplexity, or add it to your mcp.json in Cursor, Windsurf or Gemini.

  2. 2one click

    Approve

    Sign in with your AdSpends account in the OAuth window. Code editors use an API key from app.adspends.com/mcp instead.

  3. 3live data

    Ask

    Ask in plain language. Your AI picks the right AdSpends tools, pulls the data and writes the analysis.

Supported AI clients

One MCP server for every AI you use

Chat apps sign in with OAuth. Code editors and CLIs use an API key from app.adspends.com/mcp.

Connect AdSpends to Claude

OAuth

Web, desktop and mobile. Add a custom connector and approve access.

  1. In Claude, open Settings → Connectors → Add custom connector.
  2. Paste the server URL: https://app.adspends.com/api/mcp
  3. Click Connect and approve access in the AdSpends login window (OAuth).
  4. Enable the AdSpends connector in a chat via the tools menu.

Connect AdSpends to ChatGPT

OAuth

Enable Developer mode, then add AdSpends as a custom MCP connector.

  1. In ChatGPT, enable Settings → Connectors → Advanced → Developer mode.
  2. Create a connector with the MCP server URL: https://app.adspends.com/api/mcp
  3. Choose OAuth authentication and approve access in the AdSpends login window.
  4. ChatGPT uses the search and fetch tools automatically; the full tool set is available in developer mode.

Connect AdSpends to Perplexity

OAuth

Add the MCP URL under Settings → Connectors.

  1. In Perplexity, open Settings → Connectors → Add connector.
  2. Paste the MCP server URL: https://app.adspends.com/api/mcp
  3. Complete the OAuth sign-in in the AdSpends login window.

Connect AdSpends to Cursor

API key

Add the server to .cursor/mcp.json with an API key.

  1. Create an API key at app.adspends.com/mcp.
  2. Add this to .cursor/mcp.json in your project (or ~/.cursor/mcp.json) and replace the placeholder with your key.
  3. Reload Cursor. The AdSpends tools appear in the MCP panel.
JSON
{
  "mcpServers": {
    "adspends": {
      "url": "https://app.adspends.com/api/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_API_KEY>"
      }
    }
  }
}

Connect AdSpends to Claude Code

API key

One command in the terminal. Also works headless in scripts and cron jobs.

  1. Create an API key at app.adspends.com/mcp.
  2. Run this command and replace the placeholder with your key.
  3. Run /mcp inside Claude Code to check the connection.
Terminal
claude mcp add --transport http adspends https://app.adspends.com/api/mcp \
  --header "Authorization: Bearer <YOUR_API_KEY>"

Connect AdSpends to GitHub Copilot

API key

Add an HTTP MCP server from the VS Code Command Palette.

  1. In VS Code, open the Command Palette and run “MCP: Add Server…”.
  2. Choose HTTP and paste the server URL: https://app.adspends.com/api/mcp
  3. Add an Authorization header with “Bearer <YOUR_API_KEY>” (create the key at app.adspends.com/mcp).
  4. The AdSpends tools appear in Copilot Chat agent mode.

Connect AdSpends to Gemini CLI

API key

Add the server to the Gemini CLI settings file.

  1. Create an API key at app.adspends.com/mcp.
  2. Add this to ~/.gemini/settings.json and replace the placeholder with your key.
  3. Restart the Gemini CLI and run /mcp to verify the connection.
JSON
{
  "mcpServers": {
    "adspends": {
      "httpUrl": "https://app.adspends.com/api/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_API_KEY>"
      }
    }
  }
}

Connect AdSpends to Windsurf

API key

Add a Cascade MCP server in the Windsurf settings.

  1. Create an API key at app.adspends.com/mcp.
  2. Open Windsurf Settings → Cascade → MCP servers → Add server and use this config.
  3. Refresh the server list. The AdSpends tools appear in Cascade.
JSON
{
  "mcpServers": {
    "adspends": {
      "serverUrl": "https://app.adspends.com/api/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_API_KEY>"
      }
    }
  }
}

Real workflows

What you can do with the AdSpends MCP server

Six workflows you can run in minutes: the prompt, the tools the AI calls, the credits it costs and what comes back.

eCom brands & media buyers

Competitor teardown in one prompt

Analyze my competitor nordglow.de: how much are they spending on Meta, which ads are scaling, and which landing pages get the traffic?
AdSpends tools:search_advertisers· 3 crget_brand_overview· 5 crget_advertiser_ads· 10 cr
  • NordGlow (Skincare, DE): est. €412k Meta ad spend in the last 90 days, 214 active ads.
  • Top ad: a UGC video “Dermatologist reacts to my routine”, est. €58k, running for 41 days.
  • About 60% of the spend goes to one advertorial (/routine-guide); the rest goes to product pages.
  • Angle shift: 7 of the 10 newest ads lead with “barrier repair” instead of “anti-aging”.
Total: 18 credits · search_advertisers (query “nordglow”, 3 results), get_brand_overview (details, top ads, top landing pages), get_advertiser_ads (sorted by ad spend, 10 results)

Creative strategists

Find winning hooks in your niche

Find the 20 highest-spend sleep supplement ads in Germany right now and group their hooks into 5 patterns I can test next week.
AdSpends tools:list_niches· freesearch_ads· 20 cr
  • 1. “Doctor warns” authority hook: 6 ads, est. €190k combined.
  • 2. “3 a.m. wake-up” problem callout: 5 ads, est. €142k.
  • 3. Ingredient comparison vs. melatonin: 4 ads, est. €96k.
  • 4. Founder story with before/after sleep data: 3 ads, est. €71k.
  • 5. “Swiss study” proof hook: 2 ads, est. €44k. Suggested test: patterns 2 and 3 as static ads.
Total: 20 credits · list_niches (resolve “sleep supplements”), search_ads (niche + market DE, sorted by ad spend, 20 results)

Direct response marketers & publishers

Reverse-engineer scaling advertorials

Show me joint-pain advertorials that started getting spend in the last 30 days, then break down the structure of the top one.
AdSpends tools:search_landing_pages· 10 crget_landing_page· 3 cr
  • 10 advertorials found. Top: “Why Germans over 50 are skipping knee injections”, est. €96k behind 38 ads.
  • Structure: news-style headline → personal story lead → mechanism reveal (collagen type II) → doctor quote → 3 testimonials → 50% launch offer → 60-day guarantee.
  • The mechanism section runs about 900 words; the offer appears only after 70% of the page.
Total: 13 credits · search_landing_pages (keyword “joint pain”, advertorials, 10 results), get_landing_page (full page text + linked ads)

Teams using AdSpends boards

Weekly digest of your swipe file

Summarize what’s new on my “Q4 Hooks” board and tell me which saved ads gained the most spend this week.
AdSpends tools:list_saved_boards· 1 crget_saved_board_ads· 25 crget_board_statistics· 1 cr
  • “Q4 Hooks”: 25 ads from 11 brands, 6 added this week.
  • Biggest gainers: two Black Friday countdown statics (+€18k and +€12k est. spend).
  • 3 saved ads were switched off, so their hooks probably stopped working.
Total: 27 credits · list_saved_boards (find “Q4 Hooks”), get_saved_board_ads (25 saved ads), get_board_statistics (board totals)

Agencies

Pitch prep and client reporting

I’m pitching a pet food brand tomorrow. Compare the top 5 pet food advertisers in DACH by Meta spend and list 3 opportunities my prospect is missing.
AdSpends tools:list_categories· freesearch_advertisers· 5 crget_brand_overview· 25 cr
  • Top 5 by est. 90-day spend: €620k, €410k, €275k, €190k and €88k. Your prospect ranks #5.
  • Opportunity 1: the top 3 all run advertorials; the prospect only sends traffic to product pages.
  • Opportunity 2: no video ads, while 64% of competitor spend goes to video.
  • Opportunity 3: no subscription offer in the ads; 4 of 5 competitors push “save 15% with a subscription”.
Total: 30 credits · list_categories (resolve “pet food”), search_advertisers (category pet food, 5 results), get_brand_overview (× 5 brands)

Growth teams & developers

Automated monitoring with Claude Code, n8n or the REST API

Every Monday at 8:00: check my tracked brands, list new ads with more than €5k estimated spend, and post the summary to Slack.
AdSpends tools:list_tracked_brands· 1 crget_advertiser_ads· 80 cr
  • Weekly report: 8 tracked brands, 23 new ads, 5 above €5k est. spend.
  • Two competitors launched the same “price lock” angle on the same day.
  • The summary goes to #competitor-watch every Monday. No dashboard needed.
Same data via the REST API (e.g. in an n8n HTTP node)
curl -H "Authorization: Bearer <YOUR_API_KEY>" \
  "https://app.adspends.com/api/v1/ads?query=skincare&limit=5"
Total: 81 credits · list_tracked_brands (your tracked-brand lists), get_advertiser_ads (10 newest ads × 8 brands)

Illustrative examples: brand names and figures are made up. Your AI answers with live AdSpends data.

Tool reference

All 22 AdSpends MCP tools

Generated from the live OpenAPI spec. Credits are charged per call; errors are free. Your AI calls get_usage (free) to check your balance first.

ToolWhat it doesCreditsREST endpoint
Ads
search_adsSearch and browse adsResponse: ≈ 9 tokens + ≈ 290 per result1 credit per returned result (min 1, max 50)GET /ads
get_adGet one ad by idResponse: ≈ 940 tokens2 creditsGET /ads/{adId}
Brands
search_advertisersSearch advertisers by nameResponse: ≈ 11 tokens + ≈ 86 per result1 credit per returned result (min 1, max 50)GET /advertisers
get_advertiserGet one advertiser by idResponse: ≈ 130 tokens1 creditGET /advertisers/{advertiserId}
get_brand_overviewBrand overview: details, top ads and top landing pagesResponse: ≈ 2,400 tokens5 creditsGET /advertisers/{advertiserId}/overview
get_advertiser_adsList ads of one advertiserResponse: ≈ 3 tokens + ≈ 290 per result1 credit per returned result (min 1, max 50)GET /advertisers/{advertiserId}/ads
Landing pages
search_landing_pagesSearch and browse landing pagesResponse: ≈ 11 tokens + ≈ 160 per result1 credit per returned result (min 1, max 50)GET /landing-pages
get_landing_pageGet one landing page by idResponse: ≈ 3,200 tokens3 creditsGET /landing-pages/{landingPageId}
Markets, categories & niches
list_marketsList markets (valid marketId values)Response: ≈ 940 tokensfreeGET /markets
list_categoriesList advertiser categories (valid categoryId values)Response: ≈ 3,000 tokensfreeGET /categories
list_nichesList niches (valid niche values)Response: ≈ 1,900 tokensfreeGET /niches
Your saved data
list_tracked_brandsList your tracked-brand listsResponse: ≈ 9 tokens + ≈ 150 per result1 creditGET /tracking/brands
list_saved_boardsList your saved ad boardsResponse: ≈ 9 tokens + ≈ 33 per result1 creditGET /boards
get_saved_board_adsList ads on one saved boardResponse: ≈ 12 tokens + ≈ 290 per result1 credit per returned result (min 1, max 50)GET /boards/{boardId}/ads
list_saved_foldersList your saved folders of one kindResponse: ≈ 10 tokens + ≈ 37 per result1 creditGET /saved/folders
list_saved_collectionsList your saved collectionsResponse: ≈ 11 tokens + ≈ 35 per result1 creditGET /saved/collections
list_saved_tagsList your private tagsResponse: ≈ 9 tokens + ≈ 35 per result1 creditGET /saved/tags
get_saved_itemsList saved ads, landing pages or brands of a folder, collection or tagResponse: ≈ 13 tokens + ≈ 290 per result1 credit per returned result (min 1, max 50)GET /saved/items
get_board_statisticsAggregated statistics for one saved boardResponse: ≈ 49 tokens1 creditGET /boards/{boardId}/statistics
Account
get_usageYour API credit balance and costsResponse: ≈ 510 tokensfreeGET /usage
ChatGPT connector
searchChatGPT-connector alias of search_ads: returns id, title and url per ad.1 credit per returned result (min 1, max 10)MCP only
fetchChatGPT-connector alias of get_ad: the full ad as a document.2 creditsMCP only

For developers

Prefer code? Use the REST API

Every MCP tool except the ChatGPT aliases has a matching REST endpoint with the same credits. Authenticate with your API key as a Bearer token. Each response returns your remaining credits in the X-Credits-* headers.

  • MCP: Best for chat and agents: Claude, ChatGPT, Cursor, Claude Code.
  • REST: Best for scripts, dashboards, n8n, Make and data pipelines.
  • Web app: Best for visual browsing, boards and team collaboration.

Reference: API docs · openapi.json

REST API base URL
https://app.adspends.com/api/v1
curl
curl -H "Authorization: Bearer <YOUR_API_KEY>" \
  "https://app.adspends.com/api/v1/advertisers?query=snocks&limit=3"

AdSpends MCP vs. the Meta Ad Library API

The Meta Ad Library API is free, but it needs a developer account and identity verification, and it has no spend data for commercial ads. AdSpends adds the analysis layer and plugs straight into your AI.

FeatureMeta Ad LibraryAdSpends
Ad spendSpend ranges only for political and social issue ads✅ Estimated EUR spend for every Meta ad delivered in the EU
SetupMeta developer app, identity verification, access tokens✅ Paste one URL and approve with OAuth
Works in Claude, ChatGPT & CursorOnly if you build and host your own MCP server✅ Native MCP server for 8+ AI clients
Landing pagesLink URLs only✅ Advertorials and sales pages with extracted text, linked to their ads
Brand-level analysisPer-ad results, aggregate them yourself✅ Brand overview with spend, top ads and top landing pages in one call
Niches & categoriesKeyword and page search only✅ Filter by market, category and niche
Your researchNot available✅ Your saved boards, folders, tags and tracked brands

Pricing, security & troubleshooting

API credits are included in every plan

The free trial includes 100 credits. Paid plans include 2,000 to 30,000 credits per month, and you can buy credit packs on top. List tools cost 1 credit per returned result, detail tools 1 to 5 credits, and lookups like markets, niches and usage are free.

See plans & pricing →

Read-only by design

The AdSpends MCP server only reads data. It never edits your boards, never posts anything and never triggers scraping.

OAuth or revocable API keys

Chat apps sign in through OAuth, so no password ends up in a chat. API keys can be revoked at any time at app.adspends.com/mcp.

Your data stays yours

Boards, folders, tags and tracked brands are only visible to your own account. Everything else is public Meta Ad Library data.

My AI says it has no AdSpends tools.

Enable the AdSpends connector in the chat’s tools menu, or reload your editor after editing mcp.json.

Requests fail with 401 or 403.

Check that the Authorization header is “Bearer <YOUR_API_KEY>” and that your plan or trial is active.

The AI says I’m out of credits.

Ask it to call get_usage (free). Credits reset monthly, and credit packs top you up instantly.

FAQ

What is the AdSpends MCP server?

The AdSpends MCP server is a Model Context Protocol endpoint that gives AI assistants read access to the AdSpends Meta ads database. Once connected, Claude, ChatGPT, Cursor, Perplexity, Gemini, GitHub Copilot and Windsurf can search Facebook and Instagram ads, advertisers, advertorials and sales pages, and rank them by estimated ad spend derived from Meta’s EU transparency data. Key facts: MCP server URL: https://app.adspends.com/api/mcp. 22 tools across ads, brands, landing pages, markets and niches, your saved data and account usage. Read-only: the server never changes your data and never triggers scraping. Ad spend is an estimate in EUR, calculated from EU reach at a €30 CPM. Every plan includes API credits (2,000 to 30,000 per month); the free trial includes 100. The same data is available as a REST API at https://app.adspends.com/api/v1.

How do I connect the Meta Ad Library to Claude?

In Claude, open Settings → Connectors → Add custom connector, paste https://app.adspends.com/api/mcp and approve access with your AdSpends account. Claude can then search Meta ads, brands and landing pages with ad spend data in any chat.

Does the AdSpends MCP server work with ChatGPT?

Yes. Enable Developer mode under Settings → Connectors → Advanced in ChatGPT, then create a connector with the AdSpends MCP URL and OAuth. ChatGPT uses the search and fetch tools by default; developer mode unlocks all tools.

Which AI tools are supported?

Claude (web, desktop and Claude Code), ChatGPT, Perplexity, Cursor, GitHub Copilot in VS Code, Gemini CLI and Windsurf. Any MCP client that supports remote HTTP servers with OAuth or a Bearer token works.

What data can my AI access through the MCP server?

Meta ads with creatives, copy, run time and estimated spend; advertisers with brand overviews; advertorials and sales pages with extracted page text; markets, categories and niches; and your own saved boards, folders, tags and tracked brands.

How accurate is the ad spend data?

All spend figures are estimates in EUR. AdSpends calculates them from the EU reach that Meta must publish under the Digital Services Act, using a €30 CPM. They are best used to compare brands, ads and landing pages with each other, not as exact invoices.

Which countries does AdSpends cover?

AdSpends covers Meta ads delivered in the EU, because the reach data comes from Meta’s EU transparency obligations. Coverage is deepest in Germany and the DACH region.

How much does the AdSpends MCP server cost?

There is no separate charge for the MCP server. Every AdSpends plan includes API credits (2,000 to 30,000 per month) and the free trial includes 100. List tools cost 1 credit per result, detail tools 1 to 5 credits, and taxonomy and usage lookups are free.

Is it safe to connect AdSpends to my AI?

Yes. The server is read-only, chat apps connect through OAuth, and API keys can be revoked at any time. Your saved boards and tracked brands are only visible to your own account.

What is the difference between the MCP server and the REST API?

Both expose the same data and credits. The MCP server is for AI assistants and agents; the REST API at https://app.adspends.com/api/v1 is for scripts, automations and dashboards, with an OpenAPI spec at https://app.adspends.com/api/v1/openapi.json.

Can I run AdSpends research on a schedule?

Yes. Use Claude Code in headless mode, an n8n or Make workflow, or any cron job that calls the REST API. A common setup is a weekly competitor report posted to Slack.

Oleg Glanz

About the author

Oleg Glanz

Oleg is the founder and managing director of AdSpends (Publishing Media GmbH & Co. KG, Germany). He built AdSpends to show which competitor ads, advertorials and landing pages actually get budget on Meta, based on the reach data that the EU requires Meta to publish.

LinkedIn →

“Prefer clicking? The same search works in the AdSpends web app.”

Give your AI real Meta ads data

Start the free trial, paste one URL and ask your first question in under a minute.

Start free trial