AgentResolver MCP server
Commit Nov 16, 2026Monitors API, MCP, and x402 compatibility to ensure AI agents can discover, call, and pay for services.
Also known as waxsway/agentresolver, agentresolver
- Stars
- 0
- Tools
- —
- Downloads, 30 days
- —
- Latest version
- 0.2.4
- Last commit
- Nov 16, 2026
- License
- MIT
What is AgentResolver MCP?
AgentResolver monitors API, MCP, and x402 compatibility to ensure AI agents can discover, call, and pay for services.
Summary generated from the project's README
About AgentResolver MCP server
Ensuring that AI agents can successfully discover, call, and pay for services is the primary function of AgentResolver. It provides continuous monitoring for API, MCP, and x402 compatibility to prevent production failures that standard uptime monitors often miss. Developers can use the free public-service snapshot via GET or POST requests to check machine-facing discovery, or they can implement a managed monitoring schedule for $19 USDC every 30 days to receive hourly updates and a durable public status record.
For buyers of autonomous services, the Guard feature acts as a pre-sign safety check for x402 payments. It performs a bounded HTTPS preflight to validate the live payment contract, checking details like the quoted USDC amount, the recipient address, and the network before a wallet signs a transaction. This tool is designed to be non-custodial, meaning it never receives private keys or authorizes spending on its own. Users can also integrate it as an agent skill using npx or run free recurring checks through GitHub Actions.
What you can do with it
- 1Automate commerce health checksSet up a recurring schedule via GitHub Actions to monitor machine-readiness without needing an account or API keys.
- 2Prevent unauthorized autonomous spendingValidate that a live payment challenge matches the expected contract details before an agent signs a transaction.
- 3Route traffic through a reliability routerUse the x402 Reliability Router to manage facilitator traffic for existing resource servers.
Written from the project's README. Every use case is checked against a quote from it.
How to install AgentResolver MCP
Formats verified 10 Nov 2026Add to CursorAdd to GooseInstall AgentResolver MCP in Claude Code
$ claude mcp add --transport http agentresolver https://agentresolver.vercel.app/mcpInstall AgentResolver MCP in Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"agentresolver": {
"url": "https://agentresolver.vercel.app/mcp"
}
}
}Install AgentResolver MCP in VS Code
.vscode/mcp.json
{
"servers": {
"agentresolver": {
"type": "http",
"url": "https://agentresolver.vercel.app/mcp"
}
}
}Install AgentResolver MCP in VS Code CLI
$ code --add-mcp '{"name":"agentresolver","type":"http","url":"https://agentresolver.vercel.app/mcp"}'Install AgentResolver MCP in Codex
$ codex mcp add agentresolver --url https://agentresolver.vercel.app/mcpInstall AgentResolver MCP in Codex config.toml
~/.codex/config.toml
[mcp_servers.agentresolver] url = "https://agentresolver.vercel.app/mcp"
Install AgentResolver MCP in Gemini CLI
$ gemini mcp add -s user --transport http agentresolver https://agentresolver.vercel.app/mcpInstall AgentResolver MCP in Gemini settings.json
~/.gemini/settings.json
{
"mcpServers": {
"agentresolver": {
"url": "https://agentresolver.vercel.app/mcp",
"type": "http"
}
}
}Automated checks
Ran Nov 15, 2026Full report →Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points
- Source repository linked and reachablewaxsway/agentresolver is public and reachablegithub.com/waxsway/agentresolver ↗Pass
- Package names this repository as its sourceNot checked
- Build provenance (Sigstore attestation)Not checked
- Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
- Commit or release in the last yearLast commit or release 2026-11-14github.com/waxsway/agentresolver ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNot checked
- No critical advisory in the package or its dependenciesNot checked
- No hidden instructions to the model0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNot checked
- Remote endpoint answers the MCP handshakeinitialize answered with protocol 2025-06-18 in 900 msPass
README
View on GitHubAgentResolver
Continuous API, MCP, and x402 compatibility monitoring for services used by AI agents.
AgentResolver answers a production question ordinary uptime monitors miss: can an AI agent still discover, call, and pay this service right now?
Production: https://agentresolver.vercel.app
Free browser check: https://agentresolver.vercel.app/monitor
Agent Commerce Health monitoring
Start with the free public-service snapshot at GET or POST /api/service-monitor. It checks machine-facing discovery and can optionally run a real MCP initialize + tools/list exchange and a GET-safe x402 payment-contract inspection for same-origin public endpoints.
When an operator wants AgentResolver to own the schedule, POST /api/managed-monitor-30d is $19 USDC for 30 days of recurring monitoring that targets an hourly cadence with a durable public status record. Managed monitoring stores public target URLs and health snapshots only; it does not store wallet keys, API credentials, or customer contact information.
x402 Reliability Router
Existing x402 resource servers can route their standard facilitator traffic through AgentResolver without changing the merchant payTo address:
import { HTTPFacilitatorClient } from "@x402/core/server";
const facilitator = new HTTPFacilitatorClient({
url: "https://agentresolver.vercel.app",
timeoutMs: 10_000
});
Standard facilitator surface:
GET /supported
POST /verify
POST /settle
Verification is read-only and may use an explicitly configured secondary. Settlement is submitted to exactly one upstream facilitator; ambiguous timeout/5xx outcomes are marked non-retry-safe instead of being blindly failed over. Full integration guide: x402 Reliability Router.
Free recurring monitoring with GitHub Actions
Run the public-service check on your own GitHub schedule. No AgentResolver account, API key, wallet, or secret is required.
name: Agent commerce health
on:
workflow_dispatch:
schedule:
- cron: "17 8 * * *"
jobs:
health:
runs-on: ubuntu-latest
steps:
- uses: waxsway/agentresolver/.github/actions/agent-commerce-health@main
with:
origin: https://api.example.com
# Optional deeper checks:
# mcp-endpoint: https://api.example.com/mcp
# x402-endpoint: https://api.example.com/paid
The free Action reports current machine-readiness, optional live MCP initialize + tools/list compatibility, and optional GET-safe x402 payment-contract health. The hosted managed monitor is the $19 / 30-day option when you want AgentResolver to own the recurring schedule and durable status.
Full setup: free GitHub monitoring.
Buyer-side Guard
AgentResolver Guard remains the pre-sign safety check for autonomous x402 payments. Before a target wallet signs, Guard reads the live payment challenge, validates the payment contract, and returns a fail-closed decision plus evidence the caller can bind to its own policy.
What Guard checks
Guard performs one bounded HTTPS preflight against the target and verifies the live x402 contract before the caller authorizes spend:
- endpoint reachability and HTTPS/TLS evidence
PAYMENT-REQUIREDstructure- x402 version and payment scheme
- quoted USDC amount
- optional
maxPriceUsdceiling - network
- asset
payTorecipient- optional
expectedPayToandexpectedNetworkassertions - resource/payment binding
- stable evidence fingerprints for change detection
Guard blocks private/reserved network targets, does not follow redirects, uses bounded timeouts, and fails closed when required evidence is missing or mismatched.
Guard never receives the target-payment private key, signs the target payment, authorizes the target spend, or custodies/forwards target funds. The calling runtime keeps spending authority.
Price
Canonical Guard:
POST or GET https://agentresolver.vercel.app/api/x402-payment-preflight
$0.001 USDC per successful Guard check
Base or Solana where advertised
The compatibility alias /api/payment-guard uses the same Guard product. A Coinbase-CDP-specific Base route is available at /api/cdp-payment-guard for $0.002 USDC because that rail has a higher observed settlement floor.
The intended transaction path
target request
-> target returns 402
-> AgentResolver Guard
-> validate amount / asset / network / payTo / scheme / resource binding
-> Guard returns eligible or blocked + evidence
-> caller policy decides
-> caller-owned wallet may sign the target payment
Guard is evidence for a payment decision, not permission to spend.
Install as a pre-sign hook
The official x402 client exposes onBeforePaymentCreation. Use two x402 clients: one hooked client for the target payment and one hook-free client dedicated to the separate Guard payment. This prevents recursive Guard payments.
The framework-neutral client package now exports the reusable pre-sign hook directly:
import { createAgentResolverX402GuardHook } from "agentresolver-client";
targetClient.onBeforePaymentCreation(
createAgentResolverX402GuardHook({
maxTargetPriceUsd: 0.05,
guardFetch // separate payment-enabled fetch used only for the Guard fee
})
);
Until registry publication, build packages/agentresolver-client directly from this repository. The hook fails closed and binds AgentResolver evidence back to the exact selected scheme, network, asset, amount, payee, x402 version, and resource before the x402 client can proceed to signing.
Three maintained integration guides:
- x402 core pre-sign Guard
- Coinbase CDP SDK payment Guard
- Thirdweb x402 payment Guard
- Zero-dependency wallet/proxy Guard
Install as an Agent Skill
npx skills add waxsway/agentresolver --skill agentresolver-payment-guard
Direct skill source:
https://agentresolver.vercel.app/.well-known/agent-skills/agentresolver-payment-guard/SKILL.md
The skill tells a wallet-capable agent to run Guard before each unfamiliar or changed autonomous x402 spend.
Example Guard request
Guard supports a target URL plus optional caller assertions:
{
"url": "https://target.example/api",
"method": "GET",
"maxPriceUsd": 1.00,
"expectedPayTo": "0x...",
"expectedNetwork": "eip155:8453"
}
The unpaid request returns AgentResolver's own x402 challenge. After the caller authorizes and settles the Guard fee, the response contains the observed target-payment terms, reason codes, eligibility decision, and evidence receipt.
What Guard is — and is not
Guard is a transaction-path control for x402 buyers. It is designed to answer a narrow question immediately before signing:
Does the live payment challenge still match the payment contract my agent is willing to authorize?
Guard does not claim that a provider is honest, that a wallet has a particular legal owner, or that future fulfillment is guaranteed. It validates live payment-contract evidence and caller-supplied policy assertions.
Wallet-risk, reputation, AML, and other third-party evidence can be added later through explicit evidence-provider interfaces if real users require them. They are not bundled into Guard by default.
Supporting infrastructure
AgentResolver still exposes supporting infrastructure used by integrations, testing, and open-world fallback workflows. These surfaces are not the primary product:
- free
procure/resolvecapability discovery - remote MCP control plane:
https://agentresolver.vercel.app/mcp/control - x402 settlement canaries and compatibility routes
- machine-readable manifests, OpenAPI, skills, and discovery metadata
- legacy deterministic utilities retained for compatibility and evidence
New utility endpoints, passive directory submissions, and speculative marketplace-specific builds are not product strategy. Agent Commerce Health monitoring is the primary commercial product; Guard and the reliability router remain supporting buyer-side/payment infrastructure.
Machine-readable surfaces
https://agentresolver.vercel.app/.well-known/x402
https://agentresolver.vercel.app/openapi.json
https://agentresolver.vercel.app/llms.txt
https://agentresolver.vercel.app/.well-known/agentresolver-trust.json
Security
AgentResolver is non-custodial. Never provide it with wallet private keys or seed phrases.
Please report vulnerabilities privately through this repository's GitHub Security Advisories. The production security contact is also published at:
https://agentresolver.vercel.app/.well-known/security.txt
Local development
npm install
npm run dev
License
MIT
AgentResolver MCP questions
How do I add AgentResolver MCP to Claude Code?
Run claude mcp add --transport http agentresolver https://agentresolver.vercel.app/mcp in your terminal. Add --scope user to make it available in every project.
How do I add AgentResolver MCP to Cursor?
Open ~/.cursor/mcp.json and add the server under mcpServers, using the Cursor config shown in the install section above.
What does AgentResolver monitor?
It provides continuous API, MCP, and x402 compatibility monitoring for services used by AI agents.
Is AgentResolver custodial?
No, AgentResolver is non-custodial and you should never provide it with wallet private keys or seed phrases.
How much does managed monitoring cost?
Managed monitoring costs $19 USDC for 30 days of recurring monitoring at an hourly cadence.