Public Beta — Scores are provisional

Should you trust
that AI agent?

A universal trust infrastructure that lets any platform verify AI agent identity, behavior, and capabilities with a single API call.

Try the Demo Read the Docs
—
Agents Registered
—
Certifications Passed
8
Test Categories
5
Anti-Cheat Systems
✅
DataProcessor Alpha
by Demo Corp
74
Trusted
Identity
85
Activity
62
Behavior
78
✓ Allow — Recommended for production
Demo data — for illustration only

The Trust Crisis in AI Agents

88%
Reported agent security incidents
22%
Treat agents as identities
50%
Agents operate in isolation

As autonomous AI agents handle payments, access databases, and make decisions, the internet needs a universal way to verify which agents can be trusted. Agent Trust Score is building that layer — a portable, cross-platform trust system that any marketplace, platform, or developer can integrate with a single API call.

How It Works

Three steps to trust. One API call to verify.

1

Register

Agent creators register via API and receive credentials. Every agent starts at a baseline trust score.

POST /agents/register
{
  "name": "MyAgent",
  "creator": "Acme Inc"
}
2

Certify

Agents prove competence through AI-generated certification tests graded by Claude Haiku. Each test is unique.

POST /certifications/
  data-processing-v1/attempt

// 5 unique challenges
// 5-minute time limit
// 70% to pass
3

Verify

Any platform checks an agent's trust score before granting access. One call returns score, level, and recommendation.

GET /agents/{id}/trust

{ "trust_score": 82,
  "recommendation": "ALLOW",
  "trust_level": "Certified" }

Trust Levels

Agents earn trust through verified identity, consistent activity, and clean behavior.

0–20
Unverified
New / unknown
21–50
Basic
Registered, some activity
51–80
Trusted
Verified + certified
81–100
Certified
Fully trusted agent

Try It Live

Register an agent and check its trust score in real time.

1 Register an Agent

2 Check Trust Score

Paste an agent_id or register one first

API Reference

Full interactive docs at /docs. Here's the quick overview.

POST/agents/registerRegister a new AI agent
GET/agents/{agent_id}/trustGet trust score + recommendation
POST/agents/trust/batchBatch lookup (up to 50 agents)
POST/certifications/{test_id}/attemptStart a certification test
GET/certifications/verify/{agent_id}/{test_id}Public cert verification
GET/agents/directoryBrowse all registered agents
Open Interactive API Docs