Facebook Ad Library API and competitor ads tracker

1. Build the library URL

   Search the Meta Ad Library as you normally would, with a country, an active status and a keyword, or open an advertiser's Page, and copy the URL. For Google, open the advertiser or domain in the Ads Transparency Center. For LinkedIn, use an Ad Library search URL.

2. Run the tool

   Send the URL to facebook-ads in urls, or to google-ads-transparency and linkedin-ads in startUrls, with a max_results cap.

3. Compare over time

   Store the ads with their advertiser and dates, run again next week, and diff. A model on the same key can sort the copy into offers, angles and calls to action.

This run collects up to 100 active US ads that match "meal kit":

curl https://lovelyrouter.com/v1/tools/facebook-ads/run \
  -H "Authorization: Bearer $LOVELY_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "input": {
      "urls": [{"url": "https://www.facebook.com/ads/library/?active_status=active&ad_type=all&country=US&q=meal%20kit"}],
      "count": 100
    },
    "max_results": 100
  }'

Each item in items is one ad with its creative, copy, start date and advertiser. To follow one competitor instead of a keyword, put their Facebook Page URL in urls. Several URLs in one run work too; limitPerSource caps the ads taken from each. A run still going after 60 seconds comes back as 202 with a poll path, as described in Run a tool.

The Google tool takes a Transparency Center URL for an advertiser or a domain:

curl https://lovelyrouter.com/v1/tools/google-ads-transparency/run \
  -H "Authorization: Bearer $LOVELY_API_KEY" \
  -H "content-type: application/json" \
  -d '{"input": {"startUrls": [{"url": "https://adstransparency.google.com/?region=US&domain=nike.com"}]}, "max_results": 50}'

It returns each ad's format, text, the dates it was shown and its regions.

Connect the MCP server to Claude Code, Cursor or another MCP client, then ask: "Pull the active US ads of these three brands from the Meta Ad Library and the Google Ads Transparency Center, list the ones that started in the last 30 days, and group them by offer. Set max_results on every run." The agent finds the tools with suggest_tools, reads their schemas with get_tool and runs them from your balance.

The ad library tools bill per ad returned, and website-traffic bills per site plus a per-call fee. The cards above show today's prices. Starting a run holds its maximum cost, which is the per-call fee plus the per-result fee times max_results, and the run settles to what it actually returned. A run that fails pays only for the ads delivered before it ended, so a failed run with no results costs nothing. There is no subscription: you top up a prepaid balance that also pays for models and other data tools. See pricing.