Ask a language model for keyword ideas and you get a hundred plausible phrasings, none of them measured. The hard part was never generating candidates — it was knowing which ones people actually type. Here’s how an agent gets both at once.
POST a seed term and the keyword-ideas stall returns related and long-tail keywords, each carrying search volume, CPC, competition and a search-intent label — plus how many exist in total, so an agent knows whether it is seeing the whole space or the head of it. These are terms drawn from a search index, not invented, so every candidate arrives already measured. Pay-per-call over x402, no key.
# One seed term in, measured long-tail candidates out.
curl -s -X POST https://true402.dev/api/v1/keyword-ideas \
-H 'content-type: application/json' \
-d '{"seed": "pay per call api", "limit": 25}'
# → { seed, observed, totalAvailable, returned,
# ideas: [ { keyword, searchVolume, cpc,
# competition, intent } ],
# location, language, limits: [...] }Two terms with identical volume can deserve completely different pages. Someone asking what is X wants an explainer; someone asking best X for Y wants a comparison. Intent ships on every idea, so an agent can decide what to write, not merely what to write about — which is the difference between a page that ranks and a page that matches a keyword.
An idea can come back with no volume figure. That means the index has not measured it — unmeasured, not zero demand — and the response says so in a limits[] field rather than leaving the caller to guess. A seed with nothing related returns observed: false with a note instead of an empty list, because an empty list reads as “no demand here” and that is a claim about the world, not about our data.
The three keyword endpoints chain: expand a seed here, then batch every candidate into keyword-volume in a single call to size them, then run ranked-keywords to see which a domain already wins. An unpaid request returns HTTP 402; the agent pays USDC on Base and retries. New to the protocol? What is x402? Full reference: API docs.
It calls a keyword-suggestion endpoint. true402’s keyword-ideas stall does this: POST a seed term to /api/v1/keyword-ideas and it returns related and long-tail keywords, each with search volume, CPC, competition and a search-intent label. No account, no seat, no API key — the agent pays per call over x402.
A model will happily produce a hundred plausible phrasings, and it has no way to know which ones people actually type. The output reads well and is unmeasured. This endpoint returns terms drawn from a search index with a volume attached to each, so an agent can rank real candidates rather than invented ones. The two combine well: brainstorm broadly, then measure before committing.
Intent labels what a searcher is trying to do — learn something, compare options, or buy. Two terms with identical volume can deserve completely different pages: an informational query wants an explainer, a commercial one wants a comparison. Returning intent alongside volume lets an agent decide what to write, not merely what to write about.
It means the index has not measured that keyword — unmeasured, not zero demand. The response says so explicitly in a limits[] field. Reporting it as zero would assert nobody searches the term, which is a claim about the world rather than about the data, and the service will not make it.
Ideas discovers candidates, volume measures them, and ranked-keywords shows which a domain already captures. A typical agent chain is: expand one seed into a candidate list here, batch those candidates into keyword-volume in a single call to size them, then check ranked-keywords to see which are already won and which are open.