The best ad library MCP servers for competitor research
Five MCP servers that let your agent search the public ad libraries of Meta, Google, TikTok, LinkedIn and Microsoft, compared on coverage, setup and price.
What an ad library MCP server does
Meta, Google, TikTok and LinkedIn each publish a public library of the ads advertisers run, searchable by advertiser and, on most of them, by keyword. What each library shows varies by platform and country. An ad library MCP server lets your agent search those libraries and read the ads, so you can ask what a competitor is running instead of scrolling each library by hand.
The official ad platform MCP servers do a different job. Google's Google Ads MCP connects to the Google Ads accounts you can access, and its tools query your own account data through the Google Ads API. Meta's official ads MCP server manages your own Meta ads: reporting, creating and editing campaigns, catalogs and tests. Neither lists tools for searching competitors' ads in a public library. For that, you need one of the servers below.
The servers, compared
| Server | Hosted or self-run | Libraries covered | What you need | Entry price | Best for |
|---|---|---|---|---|---|
| InsightfulPipe Ads Libraries | Hosted | Meta, Google, LinkedIn, Microsoft | An InsightfulPipe workspace | Starter $29.99/mo with an Ads Libraries limit of 400 a month | Four libraries, including Microsoft, in one hosted server |
| lionkiii/gads-transparency-mcp | Local (npx) | Google Ads Transparency Center only | Node.js 18 or later, no API keys or Google account | Free (MIT) | Free Google-only research on your own machine |
| at-dan/ad-library-mcp | Self-run (your machine or your own server) | Meta and Google, plus Instagram Reels (organic posts, not ads) | A paid third-party scraping API key, optional Gemini key | Free code (MIT), plus the scraping API's credits | Meta and Google research with AI creative analysis |
| byadsco/meta-ads-mcp | Self-hosted (Docker or Cloud Run) or local | Meta Ad Library, plus your own Meta ad accounts | A Meta app or access token, plus an Apify token for the ad library tools | Free code (MIT), plus hosting and Apify (about $0.75 per 1,000 ads) | Agencies that manage client Meta accounts and also research competitors |
| Manifold | Hosted | Meta, TikTok, LinkedIn, Google | No API keys of your own | 500 free credits per workspace, top-ups from $10, Pro from $20/mo | Pay per call ad research next to SEO, Reddit and leads data |
InsightfulPipe Ads Libraries MCP
InsightfulPipe's Ads Libraries connector is a hosted server at https://ads-libraries.insightfulmcp.com/ that covers the Meta Ad Library, Google Ads Transparency Center, LinkedIn Ad Library and Microsoft Ads Library. It has 14 actions. On Meta, your agent can search ads by keyword, find a company's Page ID and list a Page's ads. On Google, it can search advertisers, list an advertiser's ads and get ad details. On LinkedIn, it can search ads by company or keyword. On Microsoft, it can search ads and advertisers and expand an ad for impressions and targeting.
Add it with claude mcp add, gemini mcp add or codex mcp add, then click Authorize access in the browser. No ad platform account is needed, only an InsightfulPipe workspace.
InsightfulPipe lists its own limits: there are no spend estimates, impressions and targeting are only available for Microsoft ads, the active or inactive filter applies to Meta only, and Microsoft's country filter covers EU and EEA countries.
Pricing is Starter at $29.99/mo, Growth at $89.99/mo and Agency at $299.99/mo, billed monthly. Each plan sets a monthly Ads Libraries limit: 400, 4,000 and 10,000.
Best for: Teams that want Meta, Google, LinkedIn and Microsoft in one hosted server, or that already use InsightfulPipe's other connectors.
lionkiii/gads-transparency-mcp
lionkiii/gads-transparency-mcp is a free, open-source server (MIT) for the Google Ads Transparency Center. It searches advertisers by domain or brand name, lists the ads an advertiser is currently running (filtered by text, image or video), gets ad details, shows an advertiser's profile and verification status, and compares ad presence across 42 regions.
Run it with npx gads-transparency-mcp and add it to Claude Desktop, Cursor, Windsurf or another MCP client. It needs no API keys and no Google account: it uses Puppeteer to read the public Transparency Center, and downloads Chromium on first install.
The README lists its limits: about 30 requests a minute, only currently running ads, a region on every request, no spend data except for political ads, and it may break if Google changes the Transparency Center.
Best for: Free Google-only ad research on your own machine.
at-dan/ad-library-mcp
at-dan/ad-library-mcp is an open-source server (MIT) for the Meta Ad Library and Google Ads Transparency Center. Its resolve-company tool finds the right brand across both, then the Meta and Google tools list a company's ads and read each ad in full: copy, carousel cards, video and targeting on Meta, and headlines, impression ranges and regions on Google. An analyze-creatives tool uses Gemini, with your own optional API key, to describe images and videos, pull out on-screen text and transcripts, and tag messaging themes. It also searches Instagram Reels, which are organic posts rather than ads.
It needs a paid API key from the third-party scraping service named in the README. Clone the repo, run npm install, add the key to .env and run npm run dev, or deploy it to your own server and add it over HTTP.
Best for: Meta and Google ad research with AI creative analysis, if you are happy to run the server and pay for the scraping API.
byadsco/meta-ads-mcp
byadsco/meta-ads-mcp is an open-source (MIT), self-hosted server for the Meta Marketing API. It is built for agencies that manage many client ad accounts, with more than 140 tools for campaigns, ad sets, creatives, audiences, reporting, comments, billing and WhatsApp Business. For competitor research, its Ad Library tools read the public Meta Ad Library by keyword or Facebook page through an Apify actor, which the README puts at about $0.75 per 1,000 ads.
Deploy it with Docker or Google Cloud Run, or run it locally over stdio for a single user. Each user signs in with Facebook Login through a Meta app you set up, or you pass an access token, and each user registers their own Apify token for the Ad Library tools. Ad Library scrapes run in the background, so the agent starts a run, waits for it and then reads the results.
Best for: Agencies that manage client Meta accounts and also want Meta Ad Library research in the same server. If you only need ad library research, the other servers here are simpler.
Manifold
Manifold is a hosted MCP server with no API keys of your own. Its ad tools read the public ad libraries of Meta, TikTok, LinkedIn and Google:
ads_search_adssearches ads by keyword on Meta, TikTok and LinkedInads_search_advertisersfinds an advertiser on Google, whose library is organised by advertiserads_get_advertiser_adslists one advertiser's ads on any of the fourads_get_adreads one ad: its copy, its offer and where it sends people
Where a library publishes numbers such as impressions or spend for an ad, Manifold passes on the library's own ranges rather than estimates. The same server also covers SEO, AI answers, social, Reddit and leads, with Search Console and Bing tools at 0 credits per call.
Manifold is pay per call. Every workspace gets 500 free credits, top-ups start at $10 and Pro starts at $20/mo. The ad library tools documentation shows what each tool returns and costs.
Limits: Manifold does not connect to your own ad accounts and does not create or edit campaigns. It does not cover Microsoft's ad library.
Best for: Pay per call ad research across Meta, TikTok, LinkedIn and Google, in the same server as your SEO, Reddit and leads research.
Which ad library MCP server should you use?
Pick by the libraries you need and whether you want to run a server yourself.
You need Microsoft's ad library, or already use InsightfulPipe. Use InsightfulPipe's Ads Libraries connector.
You only need Google and can run a local server. lionkiii/gads-transparency-mcp is free, with no API keys.
You want Meta and Google with AI creative analysis, and can run a server. at-dan/ad-library-mcp includes Gemini creative analysis. You pay for the scraping API it uses.
You manage Meta ad accounts for clients and also want competitor research. byadsco/meta-ads-mcp handles both in one self-hosted server.
You want Meta, TikTok, LinkedIn and Google in a hosted server, paid per call. Manifold covers all four, next to SEO, Reddit and leads data.
For more servers by marketing job, including paid ads, read the best MCP servers for marketing.
Run this in your agent
Connect Manifold to Claude, ChatGPT or Cursor, then paste this prompt.
Find the ads [competitor] is running on Meta and Google right now, group them by offer and hook, and tell me where each one sends people.
