Counterparty, sanctions, payment and transfer screening with reasons — for CISOs, fraud and AML teams at banks, fintechs, payment providers and digital-asset institutions. Call it from your SOC tooling, your case system or your AI agents: MCP tools for Claude, ChatGPT and Cursor, plain REST — or natively inside the Prax Sovereign AI platform, on your own hardware. One key, one URL.
Free plan: 10 credits a day, no card. Nothing you look up is shared or added to a list. Use is subject to the Terms of Service.
What an analyst, an authorization engine or an agent gets back: verdict, reasons, references, timestamps. Never rule logic, thresholds or vendor names.
Today's public API covers digital-asset counterparties and transactions plus name screening; traditional rails are part of the Enterprise platform below. Quotas are credits per day: lookups 1, batches 10, the alerts feed 100 per pull. The badge shows where a tool unlocks.
Who is behind an address before money moves. Attribution, sanctions hits, lists, Prax alerts naming it and why, exposure to flagged parties, verdict — as a short narrative plus structured fields for your case system. Milliseconds; live=true adds on-chain screening.
Pass a person, company or vessel name for fuzzy matching against the sanctions lists Prax tracks (names, aliases), or a wallet / contract address for an exact sanctions lookup. match / possible_match / no_match with programs, lists, linked addresses and scores — for onboarding, payments and periodic review.
Chain facts plus the risk of both counterparties for a settled transaction: flagged recipients, look-alike addresses, malicious tokens, unlimited approvals, created contracts. Built for fraud and AML investigators who need reasons, not a score.
Is it safe to send this amount of this asset from A to B? Verdict allow / review / block with reasons — sanctioned party, flagged counterparty, look-alike / address-poisoning, malicious asset. No chain call, so it fits inside an authorization or signing path.
Up to 50 counterparties or transfers per call with a per-item verdict and a summary — periodic customer re-screening, payment files, exposure sweeps.
Plain-language questions over Prax's threat intelligence: counts by risk band, chain and role, latest approved alerts, record lookup by id. For analysts and agents; lists are never enumerable.
The reviewed alerts Prax pushes to webhooks — attacker wallets, malicious contracts, funding ancestry, clones — last 24 hours, newest first, cursor paging. Pull it into your SIEM, case manager or agent runbooks.
Chains: eth, bsc, polygon, arbitrum, optimism, base, avalanche. Auth: Authorization: Bearer prax_…, X-API-Key or ?key=. Full reference: REST docs · OpenAPI · MCP endpoint https://mcp.orabrain.com/mcp.
No key to paste. Add https://mcp.orabrain.com/mcp to your agent; it sends you to the portal to sign in (or create a free account) and approve. Listed in the official MCP registry as com.orabrain/prax-data-services.
Settings → Connectors → Add custom connector → paste the URL. Claude opens the portal; sign in, approve, done.
One command, then /mcp → prax → Authenticate.
Installs the server into Cursor and starts the sign-in.
Copilot agent mode; MCP: Add Server flow runs the sign-in.
Settings → Connectors → Create (Developer mode) → paste the URL, auth OAuth.
Streamable HTTP, OAuth 2.1 with dynamic registration — or a Bearer key from the portal.
Approved apps appear under Account → Connected apps, each with its own key on your plan; disconnect any time. README badge: 
Pick where your agents live. Claude, ChatGPT, Cursor and VS Code use the sign-in above; for code and the Messages API create a key in the portal and replace prax_live_YOUR_KEY. Inside the Prax Sovereign AI platform the tools are already there.
Inside the Prax Sovereign AI platform nothing to paste: the platform's agents and apps call Data Services natively, on your own hardware, with no external model calls and no data leaving your environment. Banks, payment providers and digital-asset institutions run it on-premises or air-gapped, with the premium applications (payment fraud, investigation, source tracer, contract audit, ...) and traditional rails + digital assets under one policy engine. Talk to us: [email protected]
Banks, payment providers and digital-asset institutions run the complete Prax platform: a sovereign AI that reasons over your data on your own hardware — on-premises or air-gapped, no external model calls — with premium applications across traditional banking rails and digital assets, more chains and custom quotas. Data Services are built in; the same agents that answer your analysts call them natively.
Change plan any time from the portal. Enterprise adds private feeds, custom quotas and registry export.
Prax's own detection pipelines (contract scanning, chain monitoring, deployer and funding analysis), the sanctions lists Prax tracks (OFAC, UN and others), partner reputation data and human-reviewed alerts. Responses carry scores, labels and references — never the detection logic.
Requests are metered (key, tool, latency, size) for quotas and billing. Addresses, hashes and names you look up are not added to any list and are not shared.
No. The public API tools cover digital-asset counterparties, transactions and transfers plus sanctions name screening, which any financial institution needs. Traditional payment rails (cards, ACH, wires, SWIFT), payment-fraud decisioning and the investigation applications are part of the Enterprise platform.
It is built for your systems and agents: a pre-authorization verdict with reasons, a name-screening tool, a plain-words search, and MCP out of the box. It is not a case-management UI, and it does not execute transactions on a fork.
Every answer carries its reasons and references (list, alert, hop, timestamp), which is what examiners ask for. The output is intelligence to support your decision, not a legal determination — see the Terms.
Yes — the Enterprise platform runs on your own hardware, air-gapped if required, with no external model calls. Talk to us.