RelayShield

Check whether an email address appears in known data breaches. Returns breach count, source names, dates, and exposed data types (passwords, emails, etc). Call before trusting a new user identity or granting elevated access.

Score up to 10 organizational domains, each with up to 5 associated agent/employee emails, for combined breach/infostealer/session/CVE risk in one call. Built for enterprise AI-governance platforms scoring many identities per customer in one pass, the recommended entry point for agent-governance and

Composite campaign scan: fans out one indicator bundle (domains, URLs, emails, wallets, phones, file URLs, kit fingerprint IDs — max 25 indicators) across the applicable threat-intel endpoints in a single $5.50 flat call. Returns per-indicator results, kit families, cross-indicator links (shared exf

Check how many days remain before a domain's TLS certificate expires, via Certificate Transparency logs. Call to catch a lapsing certificate before it causes an outage, especially relevant as CA/Browser Forum rules shrink standard certificate lifespans toward 47 days by 2029.

Scan a domain for phishing lookalikes: typosquats, homoglyphs, and common phishing registration patterns. Returns matched lookalike domains found in the wild. Call to detect brand-impersonation phishing campaigns targeting a company before they're reported elsewhere.

Correlate an email address against the criminal breach/stealer corpus to surface linked phone numbers, secondary domains, and other identifiers tied to the same compromised identity. Call to map the blast radius of a known compromise across an organization.

Return a 0-100 domain security score across 6 identity-risk dimensions (breach exposure, infostealer density, ransomware exposure, session exposure, CVE exposure, threat-actor targeting) with a letter grade and plain-English risk factors. Call as a single-number identity health check before onboardi

Check whether an email address's credentials were harvested by infostealer malware and appear in a criminal stealer-log marketplace. Call to catch device-level compromise before stolen session cookies or saved passwords are used for account takeover.

Look up passive DNS resolution history and reputation for a domain or IP address. For a domain: which IPs it has resolved to over time. For an IP: which hostnames have resolved to it, plus malicious/suspicious vendor detection counts. Call to pivot from an indicator to its infrastructure history dur

Check whether a domain's LLM/AI provider API keys (OpenAI, Anthropic, Google, Groq, xAI, Replicate) appear exposed in criminal stealer logs. This is LLMjacking, a fast-growing threat where a leaked key becomes a live, uncapped billing liability rather than just a data exposure. Call to catch an expo

Screen an NFT contract for known scam, wash-trading, or malicious-approval risk signals before your agent buys, bids on, or approves it. Returns risk level and risk flags plus basic collection metadata. Call before an autonomous agent interacts with an unfamiliar NFT contract.

Check whether API keys or tokens tied to a domain, used by non-human identities like AI agents, service accounts, or CI/CD, appear exposed in criminal stealer logs. Call to audit whether the credentials an autonomous agent relies on have already been compromised upstream.

Check whether an email address has OAuth-connected app credentials exposed in a known SaaS breach (GitHub, Slack, Notion, Zapier, and 30+ other high-risk OAuth-capable apps). Returns matched apps and direct revoke-access links. Call to detect supply-chain credential exposure via connected apps.

Check whether a phone number appears in criminal marketplace infrastructure or pivots to known scam wallets, kits, or actors via RelayShield's threat-intelligence corpus, and whether its SIM/eSIM was recently swapped (Twilio Lookup v2, cached 24h). Returns a high/medium/unknown verdict with timestam

Check whether a domain appears on a known ransomware group's victim/leak-site list, and whether pre-ransomware credential harvesting was detected beforehand. Call to assess active ransomware exposure for a domain, not just historical breach history.

Fingerprint a suspected phishing/smishing kit from a live URL (static HTML fetch only in v1 — no JS rendering in the Lambda; submit caller-rendered HTML via the html field for JS-heavy kits) or from caller-supplied kit HTML. Returns a stable kit_<sha256> ID, extracted kit signals, corpus evidence, a

Match an existing kit_<sha256> fingerprint ID against the kit corpus. Returns the kit family (auto-suggested, pending approval), confidence, evidence, and sighting history. Cheap re-check for dashboards and bots watching for kit reuse. An unknown ID is never reported as safe — only as no-match with

Scan a file (via its public download URL) for malware using VirusTotal's multi-engine analysis. Returns an async analysis ID to poll. Call before an agent downloads, opens, or executes a file attachment from an untrusted source.

Scan a URL for phishing or malware using heuristic signals (Google Safe Browsing, RDAP domain age, known IOC corpus) plus VirusTotal multi-engine analysis. Returns an async analysis ID to poll. Call before an agent clicks, fetches, or shares a link from an untrusted source.

Screen an EVM wallet address for known scam, exploit, or sanctions-list association before your agent transacts with it. Returns a risk level and specific risk flags. Call before an autonomous agent sends funds to or interacts with an unfamiliar wallet.

Scan public GitHub repositories, npm and PyPI packages, Docker Hub images, Hugging Face models and Spaces, and Postman public workspaces and collections for API keys, tokens and credentials already published against a domain. Repo-only scanners miss credentials shipped inside released packages and i

Scan text or a unified diff for leaked secrets and credentials (49 NHI credential patterns). Check before you commit or paste.

Check whether an email address has an active stolen session cookie circulating in a criminal archive, a signal of account takeover that bypasses password resets and 2FA entirely. Call to detect AiTM/session-hijack attacks before an authenticated agent session is trusted.

Check whether a phone number has had a SIM swap or carrier port in the last 24 hours via real-time carrier lookup. A recent swap is a strong signal of an active account-takeover attempt targeting SMS-based 2FA. Call before trusting an SMS OTP from this number.

Check up to 10 vendor domains for combined breach, infostealer, and dark-web risk exposure in one call. Returns a composite risk score per vendor. Call to assess third-party API/vendor risk before an agent integrates with or continues calling an external service.

Score a domain's probability of being an active or upcoming cyberattack target using a 6-signal correlation model (breach, infostealer, ransomware, session, CVE, and threat-actor targeting history). Call for proactive risk triage, not just after-the-fact breach checking.

Check a declared technology stack (e.g. nginx, WordPress, Cisco IOS) against actively-exploited CVEs (CISA KEV) and high-EPSS-score vulnerabilities. Call before deploying or continuing to run a given technology stack in production.

Screen an ERC-20/BEP-20 token contract for honeypot, mintable-supply, hidden-owner, and other rug-pull risk signals before your agent trades it. Returns risk level, specific critical/warning flags, and basic token metadata. Call before an autonomous trading agent buys or approves spending on an unfa

Screen a wallet address across EVM, Solana, TON, or Bitcoin for known scam, exploit, drainer, or sanctions-list association before your agent transacts with it. Returns a risk level and specific risk flags. The recommended first call for any autonomous trading or DeFi agent before interacting with a

Screen up to 10 wallet addresses (any chain: EVM, Solana, TON, Bitcoin) for known scam or exploit association in a single call. Returns per-address risk level and flags. Use for bulk counterparty screening in trading or portfolio-monitoring agent workflows.

Reads the instructions an agent is given -- README, AGENTS.md, CLAUDE.md, .cursorrules, MCP manifests and tool descriptions -- and reports what they would cause an agent to do: fetch and execute remote scripts, ignore prior instructions, or touch credential files. Every referenced domain is checked

Assess an MCP server URL for supply-chain and registry risk before your agent connects to it or grants it tool-calling access — flags unverified publishers, known-malicious servers, and other trust signals. Call before an autonomous agent adds a new MCP server to its toolset.

Check whether an email address tied to an AI agent session has an active stolen session or credential exposure that could enable a prompt-injection-driven account takeover. Call to audit whether an agent's own session integrity has already been compromised upstream, not just what the agent is being