{
  "id": 11396957,
  "title": "AgentRisk: A Pay-Per-Call Risk Oracle for Autonomous Trading Agents",
  "url": "https://urgent.news/2026/10/02/agentrisk-a-pay-per-call-risk-oracle-for-autonomous-trading-agents",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-10-02T08:38:06.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/agentrisk_m2m/agentrisk-a-pay-per-call-risk-oracle-for-autonomous-trading-agents-4390"
  },
  "original_language": "en",
  "account": "AgentRisk is a pay-per-call risk oracle designed specifically for autonomous trading agents on Base. Unlike most risk tools that are built for human use, AgentRisk provides machine-readable answers via a simple HTTP call, requiring no subscription, API key, or human interaction. It cross-references GoPlus token-security data with independent on-chain checks, such as liquidity status and holder concentration, to provide a comprehensive risk assessment.\n\nAgentRisk offers signed, cacheable verdicts with each response containing a signed attestation. This attestation includes a risk score, scan ID, rulepack hash, policy version, input digest, chain, and timestamp. This allows calling agents to verify the signature offline and decide whether to cache the verdict or re-query the network, based on the timestamp's freshness. Machine-readable failure states are also provided, distinguishing between expired, revoked, replayed, and unsigned requests.\n\nThe integration surface is a REST endpoint (POST /scan) for sending contract addresses and making x402 payments, along with a MCP server for agents using Claude/MCP-compatible stacks. Additionally, the Agent Card enables other agents to discover the service without hard-coded integrations. AgentRisk has been live on Base Mainnet for months and has already processed real scans, including actual x402 payments settling on-chain. While still in its early stages, feedback from developers building M2M agent infrastructure is highly welcome to further improve the service.",
  "summary": "The problem AI trading agents on Base are increasingly autonomous — they see a token, decide, and execute, often with no human in the loop. But \"decide\" usually means trusting whatever on-chain data the agent can scrape in the few hundred milliseconds it has before the trade window closes. That's a thin foundation for judgment: honeypot contracts, unlocked liquidity, rug-pull patterns — none of…",
  "key_points": [
    "AgentRisk provides machine-readable risk assessments for autonomous trading agents on Base.",
    "Offers signed, cacheable verdicts with detailed attestation including risk score and timestamp."
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}