/shopify /agentic /ai RelayShield Order Screening is live on the Shopify App Store. Poppy brings AI agents to your storefront. RelayShield screens them. Every order checked against live threat intel (144 marketplaces, 850K+ indicators, 11M+ citations). 100 free screenings. $29/month.
25yr telecom security vet. RelayShield — SIM swap, wallet risk, breach & OAuth APIs. Devs, AI agents & DeFi traders. x402 USDC / RapidAPI. t.me/RelayShield
Top domains
11 link casts
/cybersecurity /infosec /ethereum PolinRider is back in npm. 12 hijacked releases hide a loader that reads its C2 address off Ethereum transactions. No install hooks, so disabling install scripts does nothing. Same week, bee-js shipped malicious with valid provenance and a signed app. Audit your dependency tree, not your install scripts.
/cybersecurity /infosec /metamask /base Your wallet agent moves real money. Give it a firewall. RelayShield Hosted MCP Proxy Firewall is live. Every wallet write call screened against our threat corpus + your intent binding before it reaches the wallet. Dedicated instance per tenant. $49/mo. Coinbase Wallet MCP + MetaMask.
/cybersecurity /infosec /cloud-security The address was right. The site was not. On Oct 30, a Name.com DNS hijack sent DeFi users who typed the correct address straight to a drainer kit. About $202K from 32 wallets. Read the wallet prompt before you sign.
/cybersecurity /cloud-security /infosec Northflank told customers to assume every stored env var was decrypted. A stolen AI key is a bearer key: whoever holds it can use it. Two Wiz cases: 2.5M Bedrock calls, about $100K in charges.
/cybersecurity /cloud-security /infosec Stolen AI access is now a resale business. Google and Mandiant: criminals pool stolen cloud accounts into "LLM Mixers" and sell cheap frontier access. One victim billed nearly $1M. Read more:
/cybersecurity /infosec /security VOLTA: steal, bundle, post to OpenRouter, rank by monetization value, monetize, sell. Three-model fallback when one refuses. The payoff is the outbound bundle and the inbound ranking, not just the break-in.
/cybersecurity /infosec /aiagent /digitalidentity Eon simulation: 40 percent deviation, 290 false completions, 266 prompt injections, 249 PII shares. Identity alone is not enough, intent alone is not enough, content alone is not enough. Check all three, together, at the tool call. <blog.relayshield.net/eon-intent-identity-content>
/infosec /cybersecurity Salt Security 2H 2026: nearly half of firms saw a suspected rogue agent action, but tracing is manual and stopping is not automatic. The fix is three checks at the tool call: verified identity, bound intent, runtime screening. <blog.relayshield.net/salt-rogue-agent-actions>
/ai /security /infosec /cybersecurity Forrester's 2027 predictions give shadow AI a ratio: 10 employee-built agents for every sanctioned one. They are invisible to asset inventories but visible in MCP tool traffic. Screen the tool call: tool-description drift, traversal, injection, session abuse.
/security Two npm worms, one blind spot: ChainDrop rode valid provenance from a hijacked maintainer; MALFEX rotates its package name every two days while the loader never changes. Stop detecting on names and signatures. Fingerprint the machinery. <blog.relayshield.net/two-worms-one-blind-spot>

