Endpoint-side reputation: independent verified-delivery verdicts for the full x402 catalog (~10k Base+USDC endpoints). Probes every 402 handshake, attributes settlements via facilitator-sender filtering with payTo-collision detection, flags wash/honeypots (pay-then-401), and pay-tests endpoints with real USDC — the top verdict tier requires confirmed delivery. Free rate-limited HTTP API + remote MCP: `claude mcp add --transport http x402-reputation https://x402rep.invoitech.com/mcp`. Live stats at [/api/v1/status](https://x402rep.invoitech.com/api/v1/status).

Post-quantum verification services for AI agents, self-facilitated on Base (USDC). Dilithium-2 (NIST FIPS 204) signature over any message/hash ($0.02), AI-decision attestation anchored on-chain via VAID-1 ($0.05), and a drand + PQC provenance seal for training-data/artifact provenance ($0.05). Signatures resistant to known classical & quantum attacks per NIST FIPS 204, not unbreakable. Discovery manifest at [/.well-known/x402.json](https://fractalai.net.co/.well-known/x402.json).

Independent trust & safety oracle + payment guard for x402. Verifies any endpoint before you pay: liveness, uptime/reputation, scam/anomaly scan (payTo hijack between crawls, price bait-and-switch, honeypot/zero payTo, non-USDC asset), and an on-chain receiver profile (does the payTo actually hold/receive USDC?). Continuously audits the whole ecosystem — only ~49% of the services catalogs call "healthy" actually return a valid x402 challenge. Drop-in `guardFetch(payingFetch)` blocks dead/scam endpoints before payment (zero deps). Free `/verify` + live `/status` observatory; paid `/trust` deep check. ([guard SDK](https://www.npmjs.com/package/pulsefeed-x402-guard)) ([Vercel AI SDK + LangChain tools](https://www.npmjs.com/package/pulsefeed-x402-ai-tools)) ([MCP](https://www.npmjs.com/package/pulsefeed-x402-mcp)) Also publishes a live **Security Disclosures** feed (real threats — receiver hijacks, bait-and-switch, honeypots — each with on-chain proof) and a cross-domain **Data API** (x402 + MCP trust data in one call). ([Disclosures](https://pulsefeed.dev/incidents)) ([Data API](https://pulsefeed.dev/data)) ([Observatory](https://pulsefeed.dev/status))

Buy an Agent Guild-signed, offline-verifiable trust decision before delegating work or money to an unfamiliar agent. The credential binds the capability and a bounded validity window; verification is free. Set capability in the query for the exact work being delegated.

Reputation oracle for onchain AI agents on Base. `/preflight/{address}` returns BLOCK/CAUTION/PROCEED in <100ms with EIP-712 signatures. Vet any wallet/token before x402 payments. Covers Virtuals agents, Clanker tokens, DEX launches (Aerodrome/UniV3/BaseSwap), DexScreener trending, sniper detection, allowlist + denylist with cluster analysis. Free reads + paywalled premium audits ($0.25-$25 USDC on Base). ERC-8004 registered (agentId 53953). MCP server `@raskhaaa/mainstreet-oracle`. ([GitHub](https://github.com/philpof102-svg/avisradar)) ([Agent Card](https://avisradar-production.up.railway.app/.well-known/agent.json)) ([OpenAPI](https://avisradar-production.up.railway.app/api/agent/openapi.json))

5 AI agent verification endpoints (NoLeak, MemGuard, RiskOracle, SecureExec, FlowCore) on Base Mainnet. $0.01-$0.02 USDC per call via x402.

Pre-action checkpoint API for agents. Returns `allow`, `retry`, or `block` plus a deterministic receipt before downstream writes, sends, or other actions. Live docs and x402 inventory are published on the public host. ([GitHub](https://github.com/LarryLemonBot/boundary-guard))

Open-source governance and delegation layer for x402. Provides cryptographic agent identity, scoped delegation with spending caps, rotation-aware DID verification, and signed receipts with per-condition attestation. Apache 2.0.

Free web-based scanner that validates x402 implementations on any URL (HTTP 402 with quoted price headers, payment-required surface, well-known/x402 discovery, agent-readable pricing) alongside ~70 other agent-readiness checks. Per-check remediation hints linked to a Knowledge Base. Also available as a [CLI](https://github.com/Agentic-Adventures/agentgrade-cli), [MCP server](https://agentgrade.com/mcp), and [GitHub Action](https://github.com/marketplace/actions/agentgrade-scan). Public leaderboard at [agentgrade.com/leaderboard/](https://agentgrade.com/leaderboard/).

Rail-agnostic spend-control policy layer for AI agents. Same policy engine (daily/weekly/monthly caps, categories, schedule, human-in-the-loop approval) governs x402, Stripe, and prepaid wallets. Non-custodial x402 facilitator-ready with USDC ↔ USD reconciliation, tx-hash dedup, and stablecoin de-peg protection. ([Docs](https://letagentpay.com)) ([PyPI](https://pypi.org/project/letagentpay/)) ([npm](https://www.npmjs.com/package/letagentpay))

Counterparty trust/risk scoring for x402 payments; explainable score, risk band, and reasons. $0.001 USDC on Base Sepolia (testnet). ([source](https://github.com/notifuturo/vouch)) ([/.well-known/x402](https://vouch.futuronoti.workers.dev/.well-known/x402))

Pre-signing hardening middleware for agent x402 payments: PII detection and redaction over payment metadata, spending-policy limits, and replay detection — all run before the payment is signed, fail-closed. Backed by a labelled corpus and a published study. `pip install presidio-hardened-x402`. ([PyPI](https://pypi.org/project/presidio-hardened-x402/)) ([arXiv](https://arxiv.org/abs/2604.11430))

Pre-transaction screening for agent payments. The `frisk-screen` SDK (TypeScript + Python, MIT, no runtime deps) returns an `allow`/`review`/`block` verdict before an agent pays an x402 seller, checking for the V2 `payTo` swap, spending-policy violations, plaintext-HTTP quotes, and malformed or blocklisted counterparties. Lite mode runs offline with no API key; a hosted reputation tier is in early access. ([npm](https://www.npmjs.com/package/frisk-screen)) ([PyPI](https://pypi.org/project/frisk-screen/))

Open-source cryptographic receipt layer for AI agent actions. Ed25519-signed, RFC 8785 canonicalized, content-addressed `action_ref` with hash-chained audit trails. Cited as third independent issuer in x402 section 5 (14/14 conformance verdicts). Normative implementation guidance in OWASP Agentic Skills Top 10 (AST09). Verifiable offline without operator trust. ([PyPI](https://pypi.org/project/nobulex/) | [npm](https://www.npmjs.com/package/@nobulex/core))

Pre-spend firewall for autonomous AI agents. One call before a payment returns APPROVED, BLOCKED, or FLAGGED, enforcing per-transaction caps, daily totals, velocity limits (runaway-loop protection), merchant allow/block, category rules, and time windows, plus a human-in-the-loop approval queue and a tamper-evident audit log. Rail-agnostic — governs x402, AP2, and AgentKit payments. MCP + HTTP + CLI, open-source core (MIT), free to self-host. ([GitHub](https://github.com/kindrat86/sipi-bot))

OFAC sanctions screening, transaction risk scoring, and Know-Your-Agent verification before an agent pays. Screens a counterparty wallet, name, or country against the OFAC SDN list (782 crypto wallets, 19,086 SDN entries, 16 embargoed jurisdictions) and returns an allow/review/decline decision. Compatible with x402, AP2, and AgentKit. Free tier, no API key. MCP + HTTP + CLI, MIT. ([GitHub](https://github.com/kindrat86/agentmail)) ([Agent Card](https://sanctionsai.dev/.well-known/agent-card.json))

Real-time wallet risk check for AI agents: one x402-paid GET cross-references an address's active ERC20 approvals with a live exploit threat feed and returns a risk score (0-100), risk level, and revoke-first findings before the agent acts. $0.01 USDC on Base mainnet, stateless, no signup or API key. Also available as an MCP server (`npx wallettriage-mcp`, on the official MCP Registry) and an ElizaOS plugin (`plugin-wallettriage`). ([OpenAPI](https://api.wallettriage.com/openapi.yaml))

Client-side spend firewall for agents paying via x402: enforces a user's deterministic policy before the payment is signed — destination allowlist, per-request/domain/global caps, and binding checks (signed value, recipient, asset, and capability lifetime must match the 402 challenge). Mechanism-not-policy (no reputation scoring or heuristics), fail-closed, with durable cross-process spend accounting and a tamper-evident decision log. TypeScript, MIT, zero runtime dependencies, no network egress. Pre-alpha and testnet-validated (Base Sepolia); not yet for mainnet. ([npm](https://www.npmjs.com/package/x402-spendguard))

Deterministic RU/EN safety preflight for AI-agent prompts and tasks. Flags six risk families: prompt injection, secret exfiltration, wallet signing or spend, spam or impersonation, destructive operations, and credential harvesting. POST JSON {"text":"..."}; advisory signals only, no LLM, no request-body storage, no security guarantee.

Higher-value guarded-action witness pilot for consequential autonomous actions; Base x402 USDC.

Browser-rendered page-state evidence and timestamped proof for autonomous agents; Base x402 USDC.

Pre-action verification against deterministic page expectations for autonomous browser workflows; Base x402 USDC.

Pre-pay trust gate for x402 payments. Before settling a `402`, score the payee with AgentRadar and block/warn on scam or low-trust wallets — the "Stripe Radar" for agent payments. Library-agnostic `fetch` wrapper, zero runtime deps, powered by an on-chain trust oracle (ERC-8004 reputation + static analysis + scam DB). ([npm](https://www.npmjs.com/package/@agentradar/x402-trust)) | ([GitHub](https://github.com/Bichev/agentradar-integrations)) | ([API](https://api.vvpro.ai))

Cross-signal Bayesian corroboration of entity existence, fusing WHOIS, Certificate Transparency, Wayback Machine, and DNS operational maturity into a calibrated hallucination verdict. x402 pay-per-call, tiered $0.01-$0.05 USDC on Base Sepolia by route.

Liveness check for any x402 seller address on Base mainnet. Counts real EIP-3009 (transferWithAuthorization) USDC payments received by the address over the last N days, read directly from public Base chain data. Distinguishes actual x402 revenue from ordinary USDC transfers, so a wallet with large plain inflow is not mistaken for a busy x402 endpoint. Path parameter is the address to check; optional ?days=N (default 7). Free sample of the exact output shape at /sample.

Liveness check for any x402 seller address on Base mainnet. Counts real EIP-3009 (transferWithAuthorization) USDC payments received by the address over the last N days, read directly from public Base chain data. Distinguishes actual x402 revenue from ordinary USDC transfers, so a wallet with large plain inflow is not mistaken for a busy x402 endpoint. Path parameter is the address to check; optional ?days=N (default 7). Free sample of the exact output shape at /sample.

Settlement-grounded reputation for x402 agents: a 0-1000 score derived from real on-chain USDC settlement, weighted by payer standing and sybil-resistant, so endorsements and circular vouching count for nothing. Verify any counterparty free before paying via REST (`api.agentrank.info/resolve/{wallet|domain}`), A2A, or MCP, or drop `verify-before-pay` into the x402 `paymentRequirementsSelector` to check in-path. Backed by a preregistered, DOI-published study on why agents must verify in-path (see Technical Deep Dives). ([npm](https://www.npmjs.com/package/verify-before-pay) | [GitHub](https://github.com/andysalvo/agentrank) | [Agent Card](https://agentrank.info/.well-known/agent-card.json) | [MCP](https://api.agentrank.info/mcp))

Confirms an AI agent/service is who it claims to be: fuses cross-signal domain-existence verification with a real agent-card fetch and MCP handshake.