The Google Scholar API for developers

Google has reasons to keep Scholar API-free that don't apply to most of its other products. Scholar indexes content it doesn't own — university repositories, journal publishers, preprint servers — each with its own reuse terms, so Google has no clean right to hand that corpus over through a first-party API the way it can for, say, Maps. On top of that, Scholar is rate-limited hard against automated access at the page level, which is a blunter, cheaper control than building and maintaining a public API surface.

The practical upshot: "which Scholar API is official" is not a question with a right answer to look for. Every vendor in this space, including Searlo, is an unsanctioned reader of the same public pages, and the honest way to choose between them is price and reliability, not a claim to legitimacy none of them can back.

A single search rarely answers "how influential is this paper" or "what's the newest work that cites it" — that takes a second hop, and Scholar's citation graph is exactly what citedBy, citedByLink, clusterId, and citesId expose together. citedBy is the citation count itself, and citedByLink is a direct URL to Scholar's own list of citing papers — the fast path when a count and a link are all you need. clusterId groups every version of one work under one id, so a preprint, its published copy, and a conference version resolve to the same cluster instead of showing up as three separate hits. citesId is the id you pass back as cites to pull those same citing papers through Searlo's own API instead of following the link.

All four ride along on every normal search result, so there's no separate lookup to fetch them — search once, then walk the graph from whichever result you care about. q is optional the moment you're using cluster or cites: the endpoint accepts any one of the three.

The response is bibliographic, not full-text: title, authors, venue, year, a citation count, and a link to a free PDF where Scholar surfaces one — not the paper's body. includePatents and includeCitations toggle two corpus-level choices Scholar itself exposes: whether patent literature counts as a result, and whether citation-only records (works Scholar knows exist but hasn't fully indexed) are included.

Rate limiting is the other honest caveat, and it isn't specific to Searlo: Google Scholar meters aggressively per exit IP, which is true of every Scholar-scraping API on the market. The practical implication is the same one that applies to scraping Scholar directly — avoid hammering the same query in a tight loop from a single source — rather than anything particular to this endpoint.