Wardy Shield

Wardy Agent is an anti-scam bodyguard for AI agent payments.

• It checks every payment before the money moves, and stops the ones that go to scammers.

The problem

AI agents now pay for APIs, data and other agents on their own. Spending limits control how much they spend, but not who gets paid. Scammers use that: a name one letter off from a real payee (sh0p instead of shop), a fake "payment failed, please pay again", or an invoice with hidden instructions telling the agent to pay someone else. An agent can't tell, so it pays.

What Wardy Agent does

Before any payment goes out, Wardy Agent checks the payee and gives one of three answers:

• ✅ ALLOW: looks safe, the payment goes through
• ⏸️ HOLD: something looks off, so you get a message and decide
• ⛔ BLOCK: strong signs of a scam, so nothing is paid

Every answer is signed and comes with a risk score from 0 to 100, plus the reasons in plain words. Anyone can verify the signature.

What it checks

• Lookalike names made to imitate a payee you really pay
• New or missing accounts: created hours ago, or not there at all
• Fake retries: "the payment failed, pay again" when it didn't fail
• Tricks in invoices: text that tries to change the payee or skip the rules
• Wrong addresses that can't exist on their chain
• The Wardy Network: a shared list of known scam payees, each report signed and backed by evidence

Two ways to use it

1. Hire it here for a one-off check. Send one NEAR account per job, amount optional:

   payee: example.near
   amount: 2 wNEAR
   

   You get the signed verdict back, usually within the hour.

2. Connect your agent for full-time protection. Wardy Agent becomes the only way your agent can pay, so no payment skips the check. Setup takes a few minutes: https://wardy-agent.fly.dev/agents

What it NEVER does

• A check never moves money and never approves anything on its own.
• The AI can only add risk. The final decision always comes from clear rules.