For merchants

Your API needs to respond to unpaid requests with HTTP 402 and a JSON body that lists which payment options you accept. Each option spells out the chain, the asset, the amount, and where the payment should land.

Pick any of these chains

SolanaBasePolygonArbitrumOptimismAvalancheBNBSKALE

If you're using our SDK (@dexterai/x402), this is two lines of middleware. If you're building from scratch, the protocol is documented at x402.org.

Visit x402gle.com/manage/claim and enter your domain. We'll generate a one-time DNS TXT record for you to add. Once it propagates we verify ownership and link the domain to your account.

From that moment forward, every endpoint we've discovered on your host is yours to manage. Pricing labels, descriptions, and public-or-private visibility are all editable from your merchant dashboard.

Agents calling your endpoints don't just need URLs. They need to know what each endpoint does, what to send it, and what to expect back. Writing that for every endpoint by hand is tedious, and frankly merchants don't do it well.

x402gle has an AI synthesizer (Claude Opus) that reads each of your endpoints, makes a real paid call to see how it actually behaves, and writes a structured skill description: display name, purpose, parameters, example request, expected response. One click on Synthesize all at the host level and every endpoint gets a clean description in about ten seconds.

You can review and edit what the AI wrote. When you're happy, click Looks right — that marks the skill as merchant-approved, and the public listing reflects your seal.

On every endpoint's editor page there's a Test now button at the top. The first test every hour runs on us. The endpoint gets called, the response is captured, and the AI grades it. About 3 to 5 seconds, start to finish.

You see a live cinema of every step: probing, picking a chain, paying, receiving the response, AI verdict. If the test fails, the AI tells you exactly why and which field on your editor needs editing. The Show me the fix button jumps you straight there.

Want to test more than once an hour? After the free window closes, the button switches to Test now · $0.01 (your wallet). Connect any Phantom, MetaMask, or wallet-standard wallet and pay for the retest with your own USDC. We never hold your private key. Your browser signs, our facilitator settles, money comes out of your wallet.

When a test passes with a high score, the result includes a real example request and response. That's much better evidence of what your endpoint does than the AI's initial guess. Click Save as baseline and that successful call becomes the canonical example shown on your public listing and in your skill manifest.

Agents pulling your skill manifest get the real, tested example instead of an AI's best-effort. That dramatically improves their chances of calling your endpoint correctly on the first try.