Name five competitors and your agent pulls their live ads out of the Meta, Google, TikTok and LinkedIn ad libraries, then groups them by offer, hook and format — for a fraction of a cent per call.
Ten ways into the four major ad libraries, from $0.0006 a call. The agent picks; you see the price first.
| The old way | With treg | |
|---|---|---|
| What you pay for | An ad-spy subscription, or a media-buying suite at $99–$249/mo, for research you do before each launch | One prepaid balance. A Meta ad library search is $0.00188 |
| Keys | Meta's Ad Library API needs identity verification; the rest have no official access at all | One treg token. Every tool in the catalog answers to it |
| Picking a provider | Whichever spy tool you subscribed to, whatever it happens to cover | catalog get lists every provider for that library with price, measured success rate and median speed |
| Commitment | A monthly tool for research that happens in bursts around launches | No subscription. Research a launch, stop paying until the next one |
| The workflow | Scroll four ad libraries by hand, screenshot into a slide deck, lose it before the next launch | One agent run across four platforms, grouped by offer and hook |
Paste this into the same agent. It installs the CLI, signs you in and registers the tools. One line, once — already set up? Skip to step 2.
set up treg — https://treg.to/llms.txt
Find active advertisements from these five competitors across Meta, Google, TikTok and LinkedIn. Group them by offer, hook and creative format, then identify the three most common strategies.
The agent picks a way into each library. Ten endpoints cover the four platforms, at prices that are not close to each other:
| Library | Provider | Cost | Success rate | Median |
|---|---|---|---|---|
| Meta (Facebook + Instagram) | ScrapeCreators | $0.00188 / call | 100% (166 calls) | 3.1 s |
| Apify | $0.005 / ad returned | 100% (16 calls) | 12.9 s | |
| Meta, own key | free, after identity verification | not yet measured | — | |
| Google Ads Transparency | DataForSEO | $0.0006 (async batch) | not yet measured | — |
| ScrapeCreators | $0.00188 | not yet measured | — | |
| SerpApi | $0.015 | not yet measured | — | |
| TikTok ad library | ScrapeCreators | $0.00188 | 100% (16 calls) | 3.5 s |
| Apify | $0.003 | not yet measured | — | |
| tikhub | $0.001 | 7 calls, none recorded successful | — | |
| LinkedIn ads | ScrapeCreators | $0.00188 | not yet measured | — |
| tikhub | $0.004 | not yet measured | — |
The blank cells are honest: those endpoints are catalogued and priced but have not been called often enough for treg.to to have measured them. The catalog shows you that rather than a rounded number with no sample behind it.
It pulls the creative, not just the count. Ad copy, headlines, calls to action, format, the landing URL and how long each ad has been running.
It groups and reports. By offer, by hook, by format — and how long each cluster has been live, which is the closest public signal to what is working.
Read this before you build on it. These are public ad libraries, so coverage is whatever each platform chooses to publish, and it differs by platform and country. Meta's library drops commercial ads some time after they stop running, so it is a live-ish view, not an archive. Spend figures are published only for political and issue ads.
5 competitors · 4 platforms · active ads · pulled [date] BY OFFER free trial 31 ads <competitor>, <competitor>, <competitor> demo request 18 ads <competitor>, <competitor> discount 9 ads <competitor> BY HOOK problem-first 27 ads longest-running: 94 days social proof 22 ads longest-running: 61 days ... BY FORMAT video 44 · image 31 · carousel 12 THE THREE STRATEGIES 1. <strategy> — <n> ads, running longest, used by <n> of 5 competitors 2. <strategy> — ... 3. <strategy> — ... COST $[from your run]
Structure is illustrative. Ad content comes from the platforms' own public libraries, relayed unchanged.
Run on treg.to, 17 Aug 2026, against the SEO-software category. Every figure is from the Activity log.
| Field | Value |
|---|---|
| Providers considered | 3 for Meta, 3 for Google Ads Transparency, 2 each for LinkedIn and TikTok |
| Providers selected | scrapecreators on all four libraries |
| Why | Cheapest per call at $0.00188, and the only Meta option with a real measured record — 166 calls, 100% success, 3.1 s median. Apify's Meta endpoint charges per ad returned ($0.005 each), which is the wrong shape for a broad sweep |
| Total cost of the run | $0.00752 — four calls |
| Subscription cost avoided | $447/mo at list — Optmyzr $249 + Revealbot $99 + Adalysis $99 |
| Time to completion | Around 10 seconds across all four |
| Data freshness | Live at call time; ads carried their own delivery start and end dates |
| Platforms covered | 4 of 4 — Meta, Google, LinkedIn, TikTok |
| Ads retrieved | 914 Meta matches (29 returned in full) · 2,107 LinkedIn ads (24 returned in full) · ~29,500 Google ads estimated across 3 advertiser entities |
| Cost per 100 ads | roughly $0.01 at the page sizes returned |
What came back per platform. Meta returned full creative — body copy, CTA text, display format, landing domain, page name and delivery dates. LinkedIn returned the whole post text, the person who posted it and the company promoting it. Google Ads Transparency returned something the other two cannot: the same advertiser split across three registered entities — "Semrush INC" (~20,000 ads), "Semrush Inc" (~9,000) and "Semrush Inc." (~500). Search one name and you see a third of the real activity.
Two calls in this run failed and cost nothing. One sent the wrong parameter name and was rejected by
the provider (HTTP 400, $0.00). One omitted a required parameter and was refused by treg.to before it
reached the provider at all. Neither was billed.
TikTok took two providers to get right, and that is the point. The first TikTok ad-library endpoint tried returned a 405 and was not billed. The second — ScrapeCreators at $0.00188, with a measured 100% over 16 calls — returned Semrush's running TikTok ads with video URLs, first and last shown dates and estimated audience bands. That is the objection section playing out live: treg.to does not fail over for you, it shows you the alternatives and their measured records, and you switch.
The honest read of this run: all four libraries returned data, but they do not return the same data. Meta gives you creative and delivery dates; Google gives you advertiser-level ad counts; LinkedIn gives you full post copy; TikTok gives you video files and audience bands but the
spentfield came back empty. Do not promise a clean four-platform comparison table — promise four libraries in one run, which is what actually happened.
Pull every live ad from five competitors and group them by hook. What has been running for ninety days is a stronger brief than what a team guessed in a workshop.
Search the ad libraries for the phrasing you are about to build a campaign on. If four competitors are already running it, you have learned that for less than a cent.
Run the same competitor pull weekly. New ads appearing, and old ones stopping, is the earliest public signal that someone changed strategy. A quiet week costs nothing.
Meta's Ad Library API requires identity verification and returns only what you ask for in a fields parameter; the other three platforms have no comparable official developer access. So "directly" in practice means one verified Meta app plus three manual browser workflows — for research that happens in bursts. treg.to is closer to OpenRouter for agent tools than to a data vendor: one base URL, one token, many providers behind it.
The credential is injected on the server. Your agent makes the real upstream request through treg.to's
/call/ endpoint; treg.to adds the key and relays the provider's answer back verbatim. No provider key
is ever written to your machine, your repo or your agent's context. Every call is recorded and attributed
to the token that made it.
Yes. Every endpoint has an id, and calling it by id calls that provider. If you want the choice made once
for the whole team, treg org pin <capability> --provider <provider> refuses calls to any other provider
of that capability.
(Matters here. Providers into the same library return different fields — one gives you EU
reach and per-ad detail, another gives you a cheap count. Read the endpoint before you pin one.)
Yes, and it takes precedence. Register the key once and every call to that provider routes through it — those calls are never metered against your treg.to balance. Your key always wins over treg.to's. (If you have a verified Meta app, connect it: Meta's own Ad Library API costs nothing per call, and those calls will not touch your balance.)
treg.to does not silently reroute you. That is deliberate: only you know which inputs you actually hold,
so treg.to relays your request rather than rewriting it. What your agent gets instead is the information
to recover — it already knows the alternatives and their parameters, so on a 429, a 5xx or a timeout it
can try the next provider and tell you which one it switched to. Failed calls are not billed. If a call
succeeded upstream but the answer was lost coming back, an Idempotency-Key returns the stored result
without paying twice.
A Meta ad library search is $0.00188 with ScrapeCreators, or $0.005 per ad returned with Apify. Google Ads Transparency ranges from $0.0006 to $0.015 depending on provider. The exact price is shown before the call, and treg.to adds no markup. New teams start with $1.00 of free credit — enough for several hundred library searches.
The installer registers treg.to's MCP server into Claude Code, Cursor and opencode automatically. Any MCP
client that supports the authorization spec can connect with OAuth. Anything that can run a shell command —
Codex included — can use the treg CLI or plain HTTP.
$1.00 of free credit on every new team. No credit card, no provider signup. treg.to is open source (AGPL).