Cheetah Agent Security
▸ What it does
Cheetah Agent Security provides a suite of x402 pay-per-call security services for AI agents on Base: (1) prompt-injection firewall scanning untrusted text for jailbreak attempts; (2) URL/domain reputation checking via RDAP + phishing heuristics; (3) on-chain agent trust scoring based on wallet age, activity, balances, and dispute history; (4) AML/counterparty screening against OFAC sanctions, mixer exposure, and internal scam lists; (5) agent-to-agent escrow state machine. Services priced in USDC ($0.003–$0.05 per call), billed via HTTP 402 protocol, with transparent x402 v2 metadata and OpenAPI specs.
▸ How to use it
Call the endpoints via HTTPS POST/GET with JSON payloads: (1) POST /scan with {"text": "..."} to firewall untrusted input; (2) POST /url-check with {"url": "..."} to check domain/phishing risk; (3) GET /trust?agent=0x... to score a counterparty wallet's on-chain reputation; (4) GET /screen?agent=0x... for AML screening before fund transfer; (5) POST /escrow for agent deal settlement. Each call triggers a 402 Payment Required response directing to Coinbase CDP for x402 settlement. Full OpenAPI schema available at https://x402.cheetahsecurity.de/openapi.json with input/output schemas and payment terms embedded.
▸ Evidence basis
Live, functioning infrastructure with multiple strong signals: (1) fully populated OpenAPI 3.1.0 spec with all four endpoints declared, x402 payment metadata embedded, exact pricing in USDC, and detailed request/response schemas; (2) .well-known/x402.json discovery file conforming to x402 v2 specification, listing all resources with detailed bazaar extensions explaining each service's purpose and I/O; (3) consistent metadata across on-chain registration, agent-card.json, and live API docs—no contradictions; (4) coherent value proposition: agent-centric security layer addressing real agentic internet risks (prompt injection, phishing, counterparty vetting, regulatory compliance); (5) x402 integration appears complete with Coinbase CDP facilitator declared; (6) pricing is granular and reasonable ($0.003–$0.05), tied to service complexity; (7) provider identified as Blackfort Technology with domain presence (cheetahsecurity.de); (8) explicit mention of self-hosted LLM Guard, RDAP integration, escrow dispute history—suggests operational backend. Minor caveat: homepage https://cheetahsecurity.de returns only "Moltbook x402 agent services" header (minimal content), but API infrastructure is clearly real and production-ready.
▸ Risks
Modest but real risks: (1) minimal public documentation beyond API specs—no blog, case studies, or adoption metrics to verify real usage; (2) homepage is bare, reducing trust signals for non-technical evaluators; (3) security claims (prompt-injection detection, phishing heuristics, OFAC screening) are not independently auditable from the metadata alone—effective detection depends on underlying ML models and data feeds, which are opaque; (4) no published SLAs, uptime guarantees, or dispute resolution process visible; (5) escrow feature is declared but not fully documented in the OpenAPI; (6) x402 billing via Coinbase CDP is convenient but creates a centralized dependency; (7) "our dispute history" and "our scam/dispute list" imply proprietary data—no transparency on inclusion criteria or appeals process; (8) early-stage ecosystem—unclear if agents are actually integrating this service at scale or if it remains proof-of-concept.
▸ Raw data
- Chain
- base
- Token ID
- 59098
- First seen
- 7/16/2026, 12:00:44 PM