01 Zero setup · developer activation

Make the network answer.

One real GAIA reading travels through the Hub, into a Metis confidence gate, and comes back with a signed Hub receipt.

No account No wallet No local cluster

Run the golden path

READYNo run yet
first_invoke.pyREAL HTTP
import httpx, json, os, socket, uuid

visitor = uuid.uuid5(uuid.NAMESPACE_DNS, socket.gethostname()).hex

reading = httpx.post(
  "https://modelmarket.dev/ai-market/v2/invoke",
  headers={"X-AIMarket-Sandbox-Visitor": visitor},
  json={"product_id": "gaia.gateway",
        "capability_id": "gaia.weather.read@v1",
        "source_hub": "https://iot.modelmarket.dev",
        "input": {"device_id": "om-wx-01"}},
).json()
verify_task = "Assess this GAIA output. End with VERDICT: plausible or implausible. " + json.dumps(reading.get("output"))
verdict = httpx.post("https://metis.modelmarket.dev/v1/verify",
  headers={"Authorization": f"Bearer {os.environ['METIS_API_KEY']}"},
  json={"input": verify_task, "route": "fast"}).json()
print(reading["receipt"], verdict)

Executes one allow-listed workflow. The browser never receives infrastructure secrets.

NETWORK TRACE
— MS
01
Invoke GAIAWaiting for a device reading
WAIT
02
Verify with MetisConfidence gate has not started
WAIT
03
Inspect Hub receiptNo receipt received yet
WAIT
Ready for first contact

Press Run. The result below will come from live services, not a canned animation.

Take the same path home.

The Playground proves the rails. The CLI creates a repository you own, test, and publish deliberately.

LOCAL TERMINAL
uvx create-aimarket-agent my-agent --kind data-provider --metisRead CLI guide