x402AgentTools

🩺API Health Check

Probe a GET endpoint several times and report status codes and latency statistics (min/avg/max). The cheap way for an agent to verify an API — for example one found on the x402 Bazaar — responds before spending real money calling it.

🔌 Server-side tool

This tool runs on the server (it fetches third-party URLs directly and cannot run in your browser). Use the paid JSON API below — the worked examples show real server-side results and refresh automatically.

Worked examples

Check a public API

GET /api/v1/dev/api-health-check?url=https%3A%2F%2Fapi.frankfurter.dev%2Fv1%2Flatest%3Fbase%3DUSD&runs=3

Result: ✗ UNREACHABLE — 0/3 OK · avg 0ms

Unreliable endpoint demo

GET /api/v1/dev/api-health-check?url=https%3A%2F%2Fthis-domain-does-not-exist-xyz987.com%2Fapi&runs=2

Result: ✗ UNREACHABLE — 0/2 OK · avg 0ms

Machine API (x402)

$0.002 / call

This tool is also a JSON API for AI agents. Requests without payment receive 402 Payment Required plus instructions; agents pay USDC on Base via the x402 protocol — no accounts, no API keys.

GET /api/v1/dev/api-health-check?url=https%3A%2F%2Fapi.frankfurter.dev%2Fv1%2Flatest%3Fbase%3DUSD&runs=3 HTTP/1.1
Host: agenttools-hub.vercel.app

→ 402 (payment required, instructions in headers)
→ 200 (after X-PAYMENT header; JSON body below)

{
  "tool": "dev/api-health-check",
  "input": {"url":"https://api.frankfurter.dev/v1/latest?base=USD","runs":3},
  "result": { "value": 0, "answer": "✗ UNREACHABLE — 0/3 OK · avg 0ms" }
}

Agent docs: /llms.txt · OpenAPI spec · integration guide

About this tool

An agent about to depend on an API should know it responds. This probe GETs the endpoint several times and returns status codes plus min/avg/max latency — a health verdict, not vibes. Pairs with directory-search: find the API, then health-check it.

Frequently asked questions

Only GET requests?

Yes — health probes must be side-effect free. POST/PUT testing would trigger real actions on the target system.

Why might latency differ for me?

Probes run from our server's network path. Treat relative results (unstable vs stable, fast vs slow) as reliable; absolute ms depend on geography.

Related tools