AgentPassport: The SSO Identity for AI Agents

Try sending an AI agent to buy something online. It will fail. Not because the agent is not smart enough, but because 99.9% of websites have no way to authenticate a machine visitor. CAPTCHAs block them. Login forms require human credentials. Bot detection flags them. The internet was built for humans, and it actively rejects non-human traffic.

AgentPassport changes that. It is the Google SSO for AI agents — a portable, verifiable identity that any website can check in two lines of HTML. When a merchant sees an AgentPassport, they instantly know: who built the agent, who owns it, what its trust reputation is, and whether it has funds to pay.

What Is in an AgentPassport

Every AgentPassport contains a cryptographically-signed bundle of information: the agent's unique ID, the owner's verified identity, the developer or platform that built the agent, the agent's current TrustScore, its wallet address, and a set of capability declarations (what the agent is authorized to do). The passport is issued by StableFi and can be verified by anyone without calling our servers — the signature is self-validating.

TrustScore: 0 to 1000

Attached to every AgentPassport is a TrustScore — a dynamic reputation rating from 0 to 1000. Think of it as the Dun & Bradstreet credit rating for AI agents. The score is computed from six factors:

• Transaction history — volume and consistency of successful payments
• Dispute rate — percentage of transactions that resulted in disputes
• Wallet balance — proof of funds and financial stability
• Owner verification — KYB/KYC status of the agent's owner
• Behavioral consistency — predictability of agent actions over time
• Network reputation — feedback from merchants and other agents

Merchants set their own TrustScore thresholds. A high-value merchant might require a score of 800 or above. A low-risk API provider might accept any agent with a score above 200. The flexibility is in the merchant's hands.

The Misbehavior Network

One of the most powerful features of AgentPassport is the misbehavior network. When an agent behaves badly on one merchant's site — scraping excessively, initiating fraudulent transactions, violating terms of service — that merchant reports the behavior to StableFi. The agent's TrustScore drops, and every other merchant on the network sees the updated score instantly.

This creates a powerful incentive for good behavior. An agent that misbehaves on one site loses reputation everywhere. Agent owners are motivated to configure their agents responsibly because a low TrustScore means reduced access across the entire StableFi network.

Integration: Two Lines of HTML

Merchants integrate AgentPassport by adding a meta tag and a verification script to their site. That is it. When an agent visits, the script detects the passport, verifies the signature, checks the TrustScore, and signals to the merchant's backend whether to allow or deny access. No SDK installation. No API keys to manage. Just two lines of HTML.

AgentPassport is the foundation of everything StableFi does. Identity comes first — trust, payments, and commerce all build on top of it. Learn more at stablefi.ai/registry.