Limitguard is a KYB, sanctions and PEP screening API for AI agents and developers. Your agent can reach it today over plain HTTP, MCP or A2A. Pay per call in USDC with x402, or with an API key. No subscription. The sandbox is free; wallet screening has no charge.
Also building: an agent that shops for you, inside limits you set. Join the V2 waitlist →
# request
$ curl -s -X POST https://api.limitguard.ai/v1/entity/check \
-H 'Content-Type: application/json' \
-d '{"entity_name":"Acme Corp BV","country":"NL"}'
# response (trimmed)
{
"error": "Payment Required",
"price": "0.85 USDC",
"network": "eip155:8453",
"free_sandbox": "Free sandbox key, no wallet. POST here with tier=sandbox to get an lg_sandbox_ key; send it as X-API-Key for mock data at no charge (10 req/min)."
}Recorded against the live API: the price quote, a sandbox company check, a free wallet screen and the MCP tools.
Shows: an unpaid POST /v1/entity/check answering 402 with the price and network, a company check in the browser playground with a free sandbox key, a free verify_wallet screen, and the tools listed in /.well-known/mcp.json. Nothing in it is mocked up: the sandbox result is the API's own sandbox data.
POST the entity name, country, and any identifiers you have: KVK, CBE, domain, IBAN, VAT or wallet address.
POST /v1/entity/check
Limitguard queries every check your identifiers unlock, at the same time: sanctions and PEP screening, country risk, business registry, VAT/VIES, domain WHOIS, IBAN validation.
sanctions, registry, VAT, domain, IBAN, country risk
A trust score of 80 to 100 means proceed. 60 to 79 means review. 40 to 59 means enhanced due diligence. Below 40 means block. Your agent can act on the recommendation field directly; a human only needs to look at the middle range.
{"trust_score": 87, "recommendation": "proceed"}
Live call to the Limitguard API. Free, no key.
Three ways in, all live: plain HTTP, the MCP server, and an A2A endpoint for agent-to-agent workflows.
A JSON-RPC endpoint at https://api.limitguard.ai/a2a with message/send, tasks/get and tasks/cancel. Your agent reads the agent card at api.limitguard.ai/.well-known/agent-card.json and finds the skills on its own.
entity_check, risk_score and kyb_check. Each skill runs the same handler as its REST route, so the answer is the REST answer, at the REST price.
An A2A skill call is debited from an API key's prepaid balance. To pay per call in USDC with x402, call the REST route instead; x402 inside A2A is not offered. No streaming or push notifications yet.
Point any MCP-compatible agent at api.limitguard.ai/mcp with your API key. Tools are listed in /.well-known/mcp.json.
Which checks run depends on the identifiers you send.
| Check | Coverage | Runs when |
|---|---|---|
| Sanctions and PEP screening (OpenSanctions) | Global | Every call |
| Country risk (CPI / FATF) | Jurisdiction-level | Every call |
| Business registry (KVK) | Netherlands | You send a KVK number, or a name search when the country is NL and none is sent |
| Business registry (KBO / CBE) | Belgium | You send a CBE number |
| KVK annual accounts | Netherlands | You send a KVK number |
| EU VAT (VIES) | EU | You send a VAT number |
| Domain WHOIS | Any domain | You send a domain |
| IBAN validation | SEPA | You send an IBAN |
| Wallet screening (OFAC SDN + on-chain signals) | EVM and Solana | When you call verify_wallet; not charged |
Fresh checks run $0.65 to $1.50; cached repeats run $0.10 to $0.25. Pay in USDC with x402 and no signup, or use an API key with a prepaid balance you top up once.
| Endpoint | Price |
|---|---|
/v1/entity/check | $0.85 |
/v1/risk/score | $0.65 |
/v1/reputation/score | $0.65 |
/v1/kyb/check | $1.50 |
/v1/entity/deep-check | $0.75 |
/v1/mcp/verify-wallet | Free |
See every price · How Limitguard compares
Everything you get, no subscription:
USDC with x402 on Base or Solana, no signup, or an API key debited from a prepaid balance you top up once.
A free lg_sandbox_ key returns mock data, limited to 10 requests per minute per IP. The playground creates one for you automatically.
OFAC SDN match and on-chain signals on any EVM or Solana address. Not charged.
An x402 payment settles only when the call succeeds, and a prepaid debit is refunded when it does not.
Point any MCP-compatible agent at api.limitguard.ai/mcp with your API key for direct tool access: check_entity, get_risk_score, verify_wallet.
The scoring methodology behind every check is public at GET /v1/methodology, no key needed.
A bare KVK or KBO/CBE lookup, on its own, is cheap: the Dutch Chamber of Commerce's own KVK API and independent register APIs like cbeapi.be sell exactly that, register data alone, with no sanctions, PEP, VAT/VIES, WHOIS, IBAN or scored verdict attached. Limitguard's /v1/kyb/check bundles the registry lookup with sanctions screening, country risk, EU VAT/VIES, domain WHOIS and IBAN validation, then returns one scored verdict, for $1.50 a call.
Bundled Belgian KYB from an agent-native competitor costs more per call: Strale's kyb-essentials-be is $1.62 and kyb-complete-be is $2.70 per call (source: api.strale.io/.well-known/x402.json, retrieved , via data/competitors.json).
Limitguard checks the business you are about to pay, not the agent doing the paying.
Limitguard runs its own entity check on itself and publishes the result, openly and for free.
Loading the live self-check…
Found on the official MCP registry, on Glama and Smithery, and as pip install limitguard on PyPI.
Registered as agent 95545 in the ERC-8004 Identity Registry on Base.
View on BaseScanAPI, audit log and cache on Contabo (Germany). Dashboard on Hetzner (Falkenstein, Germany). Database on Supabase (AWS eu-west-1, Ireland). Cloudflare CDN/WAF (EU-US Data Privacy Framework).
Read the trust architecture guideUptime is monitored by BetterStack and published on the status page.
status.limitguard.aiLimitguard is a KYB, sanctions/PEP and entity-trust API for AI agents and developers. It checks a company against business registers (KVK for the Netherlands, KBO/CBE for Belgium), EU VAT (VIES), OpenSanctions sanctions lists, domain WHOIS and country risk (CPI/FATF). It is hosted in the EU and operated by Ambulatio Consulting B.V. (KVK 77272471), trading as Limitguard.
There is no subscription. Fresh checks cost $0.65 to $1.50 per call depending on the endpoint; cached repeats cost $0.10 to $0.25. Wallet screening has no charge. The full price table is at limitguard.ai/pricing.
No. You can pay per call in USDC over the x402 protocol on Base or Solana with no signup at all. If you prefer an account, a free API key takes one HTTP request.
The sandbox returns deterministic mock data for any input, so you can build and test an integration before spending anything. A free lg_sandbox_ key is limited to 10 requests per minute per IP; the playground creates one for you automatically.
Yes. The verify_wallet tool checks an EVM or Solana address against the OFAC SDN list and reports on-chain signals: a contract check, native and USDC balance, transaction count and first seen on Base. It is not charged.
No answer, no charge. An x402 payment settles only when the call succeeds, and a prepaid debit is refunded when it does not.
Sanctions screening and country risk scoring run on every call, everywhere. Full business-registry lookup runs for the Netherlands (KVK) and Belgium (KBO/CBE) today; other jurisdictions get sanctions and country-risk signal but not a registry lookup.
Yes. Limitguard exposes an MCP server at api.limitguard.ai/mcp for LLM agents, an A2A agent card for agent-to-agent workflows, and x402 for autonomous payment. Its discovery endpoints are free and unauthenticated, so an agent can find and evaluate it with no human step.
It runs its own entity check on itself and publishes the result at a public, free endpoint (GET /v1/self-verify), alongside the open methodology behind every score (GET /v1/methodology).
It depends on the check. The comparison pages show sourced, dated prices against other vendors; some are cheaper per single check for a narrower result, and the comparison pages say so directly.
A score of 80 to 100 recommends proceed. 60 to 79 recommends review. 40 to 59 recommends enhanced due diligence. Below 40 recommends block. An agent can act on the recommendation directly; a human typically only reviews the middle range.
The API, its audit log and its cache run on Contabo (Germany). The dashboard runs on Hetzner (Falkenstein, Germany). The database runs on Supabase (AWS eu-west-1, Ireland). Cloudflare provides the CDN and WAF (EU-US Data Privacy Framework). See the trust architecture guide.
Free sandbox in the browser, no account. Free wallet screening, no key. Live checks from $0.65 per call, no subscription.