One name, one decision

Naming agents have a routing problem. A candidate can look clean in a trademark search while every useful domain is registered. Wikipedia may already own the phrase, and Hacker News may show years of competing use.

brand-clearance screens that candidate for $0.25 USDC over x402 on Base mainnet. Send a name plus up to six TLDs. You get a risk_level, a score from 0 to 100, the evidence behind each sub-score, and a recommendation an agent can route on.

One paid response. No spreadsheet assembly.

The endpoint asks four questions: Are there active USPTO wordmark matches? Which requested domains are registered? Does Wikipedia have a matching article? How much Hacker News history does the name have?

Those answers come back under one schema, with clear, soft, moderate, or hard as the top-level risk level. Higher scores mean more adoption risk.

Call brand-clearance

The smallest request needs one field: 

curl -X POST https://x402.agentutility.ai/brand-clearance \
  -H 'Content-Type: application/json' \
  -H 'X-PAYMENT: <signed x402 payment>' \
  -d '{"name":"Prooflayer"}'

The first unpaid request returns 402 Payment Required. Sign the quoted USDC payment and retry. Then parse an abridged response shaped like this: 

{
  "name": "Prooflayer",
  "slug": "prooflayer",
  "risk_level": "soft",
  "risk_score": 32,
  "recommendation": "Usable with caution...",
  "sub_scores": {
    "trademark": 0,
    "domain": 60,
    "wikipedia": 0,
    "hacker_news": 35
  },
  "trademarks": {
    "status": "ok",
    "count": 0,
    "hits": []
  },
  "domains": {
    "checked": ["com", "ai", "dev", "io", "co"],
    "available_count": 2,
    "registered_count": 3
  }
}

Treat that payload as screening data. If trademarks.status isn't ok, the trademark leg didn't complete, and the response says so. Don't convert an unavailable check into a clean result.

Why isn't this four calls?

Four separate responses leave the calling agent to invent a scoring policy and reconcile partial failures. It must also normalize fields and preserve the evidence that produced its decision. brand-clearance gives the router one contract instead: a stable risk bucket plus the underlying hits.

And one x402 settlement buys that whole screening result.

That changes the control flow. A naming agent can keep clear, flag soft, require human review for moderate, and stop on hard. The evidence remains available when a human asks why the name was rejected.

Price the shortlist before calling. Eight candidates cost $2.00. If the agent has 80 generated names, rank them first and spend the clearance budget on finalists.

Where handles and WHOIS fit

brand-clearance doesn't check social handles. It also doesn't return a full WHOIS or RDAP history. Its domain section answers the adoption question agents usually need first: is each requested name registered, and which registrar appears when available?

Need the X handle too? Call x-handle-availability after a candidate survives clearance. Need creation dates, expiry, DNSSEC, or EPP status? Call whois-lookup on the winning domain.

If your router requires domains, a social handle, clearance, and lookalike risk in one response, choose brand-presence-report. That endpoint has the broader contract. Route by the output you need, not by a loose reading of “brand check.”

Keep the legal boundary visible

A clear result isn't a legal opinion. The response doesn't cover state registries, common-law use, foreign jurisdictions, or design marks. For commercial adoption, retain a trademark attorney and pass the endpoint's evidence into that review.