If you already let a coding agent run chores for you, competitor backlinks are a good job to hand over. You want a competitor backlinks API that returns linking sites for a domain in a structured form, a price you can cap per call, and an agent that does the set arithmetic. seogeoaeo.ai fits that shape: its Backlink Finder is exposed over REST and MCP, priced per run, and the rest is a few lines of logic.
What the endpoint returns
Backlink Finder takes one input, a public URL. A page address is reduced to its domain, because the lookup covers the whole site. The result lists the 100 strongest sites linking to that domain, strongest first. For each linking site you get:
- a link rank from 0 to 1,000
- how many links it gives the domain
- the date its first link was seen
- a spam score out of 100, with 40 or more marked likely spam
- the share of its linking pages that are followed
It also returns the 50 link texts used by the strongest sites, grouped as brand name, web address, generic words, no text or other words, plus the total number of linking sites as the data provider counts them.
Calling it
Over REST, it is one POST:
curl -X POST https://seogeoaeo.ai/api/v1/tools/backlink-finder/runs \
-H "Authorization: Bearer $SEOGEOAEO_API_KEY" \
-H "Content-Type: application/json" \
-d '{"input":{"url":"https://rival-one.example"},"maxCredits":70}'
maxCredits makes the run refuse to start if the price is above your cap, which is the setting you want on anything an agent triggers. idempotencyKey lets a retry go through without paying twice. Keys start with sga_ and spend the workspace's credits.
Over MCP, add the server at https://seogeoaeo.ai/api/mcp with a bearer header. The tool name is the slug, backlink-finder. It works in Claude Code, Cursor, VS Code, Windsurf and Codex, and agent runs cost the same as runs in the browser.
The gap logic
Ask the agent to do this, or write it yourself:
- Run Backlink Finder on each competitor and on your own domain.
- Normalise every linking site to its root domain and drop "www.".
- Drop rows with a spam score of 40 or more. Look at borderline ones by hand, since the score is an estimate.
- For each remaining site, count how many competitor lists contain it.
- Keep sites with a count of two or more that are missing from your list.
- Sort by link rank and write the result to a CSV.
Step 4 is where the value is. A site linking to several rivals already covers your category, so it is a stronger lead than one strong link to a single rival. Use the same source for every domain; mixing link indexes produces gaps that are not real.
What it costs
Each run is 70 credits. Three rivals plus your own site is 280 credits. The Indie plan gives 2,000 credits for $29 a month, Startup 5,000 for $69, Scale 22,000 for $199. Top-ups ($19 for 1,000 credits and up) only work alongside an active plan and do not expire. A run that finds no linking sites and no link texts is refunded.
If the agent should also check whether the rivals are the right ones, Competitor Finder (38 credits) lists the 30 sites sharing the most Google keywords with a domain, and Domain Overview (50 credits) gives a quick size-up with total linking sites and link rank.
Limits to tell your agent about
- 100 strongest linking sites per domain, not every link.
- Whole domains only. No per-page lookups.
- No contact details, so outreach still needs a human or another tool.
- No history and no alerts. To see change, store each run and diff later.
- Figures come from a link data provider and are estimates.
If you need complete backlink exports for very large sites, a full suite's API is a better fit. Semrush's Backlink Gap, for example, compares up to five sites at once inside its interface. For a lean pipeline that answers "who links to them and not to me", per-run pricing keeps the bill tied to what you actually run.
What the agent should not do
Do not let it draft outreach that offers payment or free products for links. Google's spam policies treat that as link spam unless the link is marked rel="sponsored" or rel="nofollow". Have it tag each lead by page type instead: directory, roundup, resource page, guest article, news, partner page. Roundups and resource pages are where replies come from.
Setup details and every tool's inputs are in the seogeoaeo.ai agent reference.