Agent-Authorized Commerce

Your agent needs
human clearance.
One API call.

The missing auth layer between human intent and agent execution. Agents request, humans approve, services verify. Calm enough for a real team, strict enough for money movement.

Get API Key — Free View API →

Human-in-the-loop by default

USDC on Base for paid plans

Readable by people and bots

Operator Layer

Approval Online

Request Book service van

Budget: $180. Agent waits. Human reviews. Nothing moves until someone actually says yes.

Pending

Approved

Agent View

POST /v1/clearances
scope: "book:service"
budget: 180 USD
status: pending

Glitch

★★★★★

Keeper patch
human approved

Glitch Keeper works better here as a stitched approval patch than a giant mascot face. Clearance still feels serious, just with some personality.

                
                
                
            

# Agent requests human approval
curl -X POST https://clearance-production-5081.up.railway.app/v1/clearances \
  -H "X-API-Key: clr_live_your_key_here" \
  -H "Content-Type: application/json" \
  -d '{
    "title": "Book flight to Austin",
    "scope": "purchase:flight",
    "budget_amount": 450.00,
    "budget_currency": "USD"
  }'
            

# pip install httpx
import httpx

resp = httpx.post(
    "https://clearance-production-5081.up.railway.app/v1/clearances",
    headers={"X-API-Key": "clr_live_your_key"},
    json={
        "title": "Book flight to Austin",
        "scope": "purchase:flight",
        "budget_amount": 450.00,
    }
)
clearance = resp.json()
# Send clearance["approval_url"] to the human
# Poll clearance["id"] until status == "approved"
# Use clearance["token"] to prove authorization
            

{
  "id": "clr_a1b2c3d4e5f6g7h8",
  "status": "pending",
  "title": "Book flight to Austin",
  "scope": "purchase:flight",
  "budget_amount": 450.00,
  "approval_url": "https://clearance-production-5081.up.railway.app/approve/clr_a1b2...",
  "expires_at": "2026-04-16T02:35:00Z"
}

// Human clicks approval_url -> Approves with one tap
// Agent polls GET /v1/clearances/{id}
// Gets back a signed JWT token proving approval
// Any service can verify: GET /v1/verify/{token}
            

Three parties. One flow.

Agent requests. Human approves. Service verifies.

Agent Requests Clearance

Your AI agent calls our API with what it wants to do, the scope, and the budget. Gets back an approval URL instantly.

AI Agent

Human Approves

The human gets a clean, mobile-friendly page showing exactly what the agent wants. One tap to approve or deny.

Human

Service Verifies

The agent receives a signed JWT token. Any service can verify it with one API call. No auth required to verify.

Any Service

Simple pricing. Card or crypto.

Start free. Scale when you need to.

Starter

Forever free

50 clearances / month
Approval pages
Token verification
API docs

Pro

$19/mo

For serious agents

1,000 clearances / month
Webhook callbacks
Priority support
Custom metadata

Scale

$49/mo

High-volume agents

10,000 clearances / month
Webhook callbacks
SLA guarantee
Bulk token ops

Pay by card through Stripe Checkout or use USDC on Base. Other assets or networks are accepted only through invoice-confirmed instructions priced to a USD-equivalent amount. Fixed-price plans verify and issue keys automatically after confirmed payment. Refunds: consulting@nauti-labs.com

For AI Agents

Built for machines. Readable by machines.

Full OpenAPI spec at /v1/docs. Every response is structured JSON. Auth via header. Stateless.

POST /v1/clearances Create approval request

GET /v1/clearances/{id} Poll for decision

GET /v1/verify/{token} Verify approval (public)

POST /v1/clearances/{id}/revoke Revoke clearance

GET /v1/usage Check credits

POST /v1/keys Create API key

Your agent needshuman clearance.One API call.