LinkCensus API and MCP docs
Give scripts and AI agents the same dated link-neighborhood candidates the website shows, through a REST API, the linkcensus command line or a local MCP server.
How it fits together
Every request uses your own API key. However you call LinkCensus, neighborhood lookups count against one monthly allowance on your account.
REST API
GET /api/v1/neighbors/{domain}
Command line
linkcensus neighbors example.com
Local MCP server
get_neighbor_candidates
5 neighborhood lookups, up to 5 rows each
One counter for every key and client on the account. Edition and domain checks do not use it.
Link-neighborhood candidates
Each row keeps its edition, ranking mode and shared linkers (among top 200 retained).
Where to start
- REST API and CLI
Authenticate with a LinkCensus API key, list editions, resolve a domain and fetch dated neighbor candidates over REST or with the linkcensus command line.
- MCP server
Connect an MCP client to https://linkcensus.com/mcp with your API key, or run the local stdio server, and call list_editions, resolve_domain and get_neighbor_candidates.
- Usage and limits
How the free agent allowance works: 5 neighborhood lookups a month with up to five rows each, shared by REST, the CLI and MCP, and separate from website lookups.
What a result is
A result is a ranked list of domains that sit near yours in a dated Common Crawl link graph, with the shared linkers behind each one where the edition retained them. Treat each row as a research lead: graph proximity is not proof that a site is a competitor, and it says nothing about traffic or revenue. The method page explains how neighbors are found.