How it works Security Docs Pricing Sign In / Get Started

Everything you need
to integrate ARIA.

Full quickstart, SDK reference, and API docs: get up and running in under 5 minutes.

Quickstart Track Modes Scope Trust Score Webhooks API Reference Examples

Quickstart

1

Create your account

Go to ariatrust.org/app, register and confirm your email. Copy your API key from the dashboard.

2

Install the SDK

bash 
npm install @ariatrust-io/aria-sdk
3

Register your agent

typescript 
import { createClient } from '@ariatrust-io/aria-sdk';

const aria = createClient({
  baseUrl: 'https://ariatrust.org',
  apiKey: process.env.ARIA_API_KEY
});

const agent = await aria.registerAgent({
  name: 'my-agent',
  scope: ['read:data', 'write:data', 'send:email']
});

// Save these: you need them to track events
console.log(agent.did);    // did:agentrust:...
console.log(agent.secret); // keep this secret
4

Track actions

typescript 
const result = await aria.track(
  agent.did,
  agent.secret,
  'read:data',
  async () => {
    return await fetchUserData(userId);
  }
);
result.insights
{
  scope: { valid: true },
  signature: { valid: true },
  rateLimit: { exceeded: false },
  trustScore:{ impact: +1 }
}
5

View in dashboard

Sign in at ariatrust.org/app to see your agent, its trust score, and every action it has taken, with full cryptographic proof.

Track Modes

ARIA supports three modes for tracking agent actions. Choose the mode that matches the risk and latency requirements of each action.

enforce

Default

Blocks until ARIA confirms the event. Use for critical actions where you need insights back.

light

Low latency

Fire-and-forget. Agent continues immediately. Use for high-frequency actions where latency matters.

gate

Human-in-loop

Requires human approval before executing. Use for destructive or high-risk actions.

typescript · enforce (default) 
const result = await aria.track(did, secret, 'read:data', fn);
// Waits ~150-300ms for ARIA response
typescript · light 
const result = await aria.track(
  did, secret, 'read:data', fn,
  { mode: 'light' }
);
// Returns immediately: ARIA records in background
typescript · gate 
import { GateDeniedException } from '@ariatrust-io/aria-sdk';

try {
  await aria.track(
    did, secret, 'delete:records', fn,
    {
      mode: 'gate',
      gate: {
        requireApproval: ['delete:*'],
        autoBlock: ['drop:*'],
        timeoutMs: 5 * 60 * 1000
      }
    }
  );
} catch (err) {
  if (err instanceof GateDeniedException) {
    console.log('Owner denied this action');
  }
}

Scope

Scope declares what your agent is allowed to do. Format: verb:resource. If an agent attempts an action outside its declared scope, ARIA records a scope violation and penalizes the trust score.

read:users Read user data
write:orders Create or update orders
send:email Send emails on behalf of users
process:payment Process payment transactions
generate:report Generate and export reports
typescript 
const agent = await aria.registerAgent({
  name: 'my-agent',
  scope: [
    'read:users',      // read user data
    'write:orders',    // create/update orders
    'send:email',      // send emails
    'process:payment', // process payments
    'generate:report'  // generate reports
  ]
});

Trust Score

Every agent has a score from 0 to 100 based on behavior over the last 30 days. The score decays toward 50 after 7 days of inactivity.

Score Changes

Successful action +1
Error outcome -1
Anomaly detected -5
Scope violation -15 to -50
Hardware conflict -20 to -40

Notes

Maximum achievable ~85
Score never reaches 100 by design
Inactivity decay toward 50
Decay starts after 7 days
Scope violations escalating
Trusted
80 – 100
Fully verified behavior
Neutral
50 – 79
Monitor for anomalies
Untrusted
0 – 49
Investigate immediately

Webhooks

Get notified in real time when something suspicious happens. Each webhook is signed with HMAC-SHA256 via the X-ARIA-Signature header so you can verify authenticity.

bash · register webhook 
curl -X POST https://ariatrust.org/v1/webhooks \
  -H "Authorization: Bearer your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://your-server.com/aria-alerts",
    "events": [
      "anomaly",
      "scope_violation",
      "trust_score_critical"
    ]
  }'
json · webhook payload 
{
  "alert": "TRUST_SCORE_CRITICAL",
  "severity": "CRITICAL",
  "agent": {
    "did": "did:agentrust:...",
    "name": "my-agent",
    "trustScore": 12
  },
  "reason": "scope_violation",
  "timestamp": "2026-05-14T10:00:00Z"
}

API Reference

Base URL: https://ariatrust.org  ·  Authentication: Authorization: Bearer <api-key>

Method Endpoint Description
GET/healthServer status
POST/v1/auth/registerCreate account
POST/v1/auth/loginSign in (sends 2FA code)
POST/v1/auth/verify-codeVerify 2FA, receive API key
POST/v1/agentsRegister agent
GET/v1/agentsList agents
GET/v1/agents/:didAgent details + trust score
GET/v1/agents/:did/secretRecover agent secret
DELETE/v1/agents/:didDelete agent
POST/v1/agents/bulk-deleteDelete multiple agents
POST/v1/eventsTrack single event
POST/v1/events/batchTrack up to 500 events
GET/v1/eventsList events
GET/v1/events/exportExport events as CSV, JSON, or OTEL (OpenTelemetry, compatible with Splunk, Datadog, CloudWatch)
POST/v1/gate/requestRequest human approval
GET/v1/gate/request/:idCheck approval status
POST/v1/gate/approve/:idApprove action
POST/v1/gate/deny/:idDeny action
GET/v1/gate/pendingList pending approvals
POST/v1/webhooksRegister webhook
GET/v1/webhooksList webhooks
DELETE/v1/webhooks/:idRemove webhook
POST/v1/api-keysCreate API key
POST/v1/api-keys/rotateRotate API key
POST/v1/witness/sourcesRegister external witness source
POST/v1/witness/confirm/:idSubmit external event count
GET/v1/witness/checksList verification checks
GET/v1/witness/agents/:didWitness summary for agent
POST/v1/temporal/anchor/:didCreate temporal anchor
GET/v1/temporal/verify/:eventIdVerify event proof
GET/v1/temporal/anchors/:didAnchor summary for agent
GET/v1/temporal/anchors/:did/listList all anchors for agent
POST/v1/zeroproof/innocenceGenerate Proof of Innocence
POST/v1/zeroproof/consistencyGenerate Proof of Consistency
POST/v1/zeroproof/limitsGenerate Proof of Limits
GET/v1/zeroproof/verify/:idVerify a proof
GET/v1/zeroproof/list/:didList proofs for agent

Examples

typescript · openai function calling 
import OpenAI from 'openai';
import { createClient } from '@ariatrust-io/aria-sdk';

const openai = new OpenAI();
const aria = createClient({ baseUrl: '...', apiKey: '...' });

const agent = await aria.registerAgent({
  name: 'openai-agent',
  scope: ['read:data', 'send:email', 'write:calendar']
});

// Wrap each tool call with ARIA
async function callTool(toolName: string, args: unknown) {
  return aria.track(
    agent.did, agent.secret, toolName,
    async () => executeTool(toolName, args),
    { mode: 'light' } // non-blocking for high frequency
  );
}
typescript · langchain tools 
import { DynamicTool } from 'langchain/tools';
import { createClient } from '@ariatrust-io/aria-sdk';

const aria = createClient({ baseUrl: '...', apiKey: '...' });
const agent = await aria.registerAgent({
  name: 'langchain-agent',
  scope: ['search:web', 'read:database', 'send:email']
});

// Wrap LangChain tools with ARIA tracking
function wrapTool(tool: DynamicTool, action: string) {
  return new DynamicTool({
    name: tool.name,
    description: tool.description,
    func: async (input: string) => {
      const result = await aria.track(
        agent.did, agent.secret, action,
        async () => tool.func(input),
        { mode: 'light' }
      );
      return result;
    }
  });
}