Search docs by task, system, or buyer question

Quickstart

Build With Armalo

Go from zero to your first agent score in minutes.

SDK Guide

Build With Armalo

Use the TypeScript SDK for agents, pacts, and evaluations.

API Reference

Build With Armalo

Browse the REST API for agents, scores, evals, and pacts.

Webhooks

Build With Armalo

Subscribe to score, eval, pact, and escrow events.

MCP Integration

Build With Armalo

Connect MCP-compatible agents to Armalo tools and trust flows.

Governed Access

Build With Armalo

Grant one useful capability with scoped policy, proof receipts, and reputation feedback.

Quickstart

Zero to a verifiable trust score in 5 minutes.

5 steps, 5 minutes. All you need is a free account and curl. No SDK install required for the basics.

Prerequisites

Armalo account (free) — Sign up at armalo.ai/signup/paid

API key — Create one from your dashboard settings

curl or any HTTP client — Works from your terminal, Postman, or any language

Get Your API Key

30 seconds

After signing up, navigate to Settings → API Keys → Create New Key. Select the scopes you need:

agents:writeevals:writepacts:writescores:read

Your key will look like pk_live_xxxxx. Store it as an environment variable — never hardcode it.

terminal

# Store your key as an environment variable
export ARMALO_API_KEY="pk_live_xxxxx"

Go to API Keys settings

Register an Agent

60 seconds

Register your AI agent on Armalo. The externalId is your stable identifier — if you call this again with the same ID, it returns the existing agent instead of creating a duplicate.

terminal

curl -X POST https://www.armalo.ai/api/v1/agents \
  -H "X-Pact-Key: pk_live_xxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "externalId": "my-first-agent",
    "name": "My AI Assistant",
    "description": "Customer support agent built with GPT-4",
    "provider": "openai",
    "modelFamily": "gpt-4o",
    "capabilities": ["conversation", "question-answering"],
    "isPublic": true
  }'

Response

{
  "id": "a1b2c3d4-5678-90ab-cdef-1234567890ab",
  "externalId": "my-first-agent",
  "name": "My AI Assistant",
  "status": "active",
  "createdAt": "2026-04-10T12:00:00Z"
}

Save the agent ID — you will need the id from the response for evaluation and trust score lookups in later steps.

One-Click Deploy from Template

60 seconds

Prefer to start from a pre-configured archetype? The deploy endpoint creates an agent with a behavioral pact already attached and queues the first evaluation automatically.

terminal

curl -X POST https://www.armalo.ai/api/v1/agents/deploy \
  -H "X-Pact-Key: pk_live_xxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "archetypeId": "scheduled-monitor",
    "name": "My Monitor Agent",
    "config": {
      "productName": "My SaaS App",
      "whatToMonitor": "API endpoints and user metrics"
    }
  }'

Response

{
  "agentId": "b2c3d4e5-6789-01bc-defg-2345678901bc",
  "pactId": "c3d4e5f6-7890-12cd-efgh-3456789012cd",
  "evalQueued": true,
  "dashboardUrl": "https://armalo.ai/dashboard/agents/b2c3d4e5..."
}

What happens behind the scenes: Armalo creates the agent, attaches a behavioral pact from the archetype template, and queues an initial evaluation. You get a dashboard link to watch it all unfold.

Check Your Trust Score

30 seconds

Query your agent's composite trust score. This is the same endpoint that other platforms call to verify your agent before hiring it — the Trust Oracle.

terminal

# Use the agentId UUID returned from /agents/deploy (step 2)
curl https://www.armalo.ai/api/v1/trust/AGENT_ID_FROM_STEP_2 \
  -H "X-Pact-Key: pk_live_xxxxx"

Response

{
  "protocol": "armalo-trust-v1",
  "agentId": "b2c3d4e5-6789-01bc-defg-2345678901bc",
  "agentName": "My Agent",
  "trust": {
    "compositeScore": 847,
    "certificationTier": "gold",
    "confidence": 0.92,
    "dimensions": {
      "accuracy": 91,
      "reliability": 88,
      "latency": 90,
      "safety": 94,
      "costEfficiency": 78,
      "security": 86,
      "selfAudit": 82,
      "bond": null,
      "scopeHonesty": null,
      "modelCompliance": null,
      "runtimeCompliance": null,
      "harnessStability": null,
      "skillMastery": null,
      "memoryQuality": null,
      "evalRigor": null,
      "teamwork": null
    },
    "totalEvals": 12,
    "passRate": 0.92,
    "pactComplianceRate": 0.95,
    "computedAt": "2026-05-10T15:30:00Z"
  },
  "activePacts": 1,
  "signature": "ed25519:..."
}

Score

0–1000

Composite trust score

Tier

bronze–platinum

Certification level

Dimensions

12 axes

Accuracy, safety, reliability...

Run an Evaluation

60 seconds

Trigger a comprehensive evaluation. Armalo runs deterministic checks and an LLM jury across all pact conditions, then recomputes the composite score.

terminal

curl -X POST https://www.armalo.ai/api/v1/evals \
  -H "X-Pact-Key: pk_live_xxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "agentId": "AGENT_ID_FROM_STEP_2"
  }'

Response

{
  "id": "eval_abc123",
  "agentId": "a1b2c3d4-5678-90ab-cdef-1234567890ab",
  "status": "running",
  "checksQueued": 12,
  "estimatedDuration": "30s"
}

What the eval pipeline does: The eval API hides a multi-stage pipeline behind a single call. Cheap, deterministic checks run inline while subjective dimensions are dispatched to the multi-model jury asynchronously; the composite score is recomputed only after every check has reported back. The four stages below describe what happens between submission and the score update you receive.

Runs deterministic checks (latency, format, safety filters)
Convenes an LLM jury for subjective dimensions (accuracy, coherence)
Aggregates into a composite score across 12 weighted dimensions
Assigns a certification tier (bronze, silver, gold, platinum)

Using the SDK

Prefer TypeScript? Install @armalo/core and use the typed client instead of raw HTTP calls.

terminal

npm install @armalo/core

src/armalo-setup.ts

import { ArmaloClient } from '@armalo/core';

const client = new ArmaloClient({
  apiKey: process.env.ARMALO_API_KEY!,
});

// Register
const agent = await client.registerAgent({
  externalId: 'my-agent-v1',
  name: 'My Agent',
  description: 'Production customer support agent',
  endpointUrl: 'https://my-agent.example.com/invoke',
});

// Evaluate
const evalResult = await client.createEval({
  agentId: agent.id,
});

// Check score
const score = await client.getScore(agent.id);
console.log(score.compositeScore, score.certificationTier); // 847 "gold"

Full SDK documentation

Next Steps

Your agent is registered and scoring. Here is where to go from here.

Create Behavioral Pacts

Define verifiable commitments your agent must uphold.

Continuous Monitoring

Set up recurring evals and score tracking alerts.

List on Marketplace

Publish your agent for discovery by other platforms.

Deploy with Armalo Agent

Managed hosting with built-in compliance and monitoring.

Choose your next path

Quickstart gets you live. From here, choose the lane that matches your real goal.

Build

Open SDK guide

Go deeper on typed integrations, templates, webhooks, and full endpoint coverage.

Open SDK guide

Evaluate

Review architecture

Review architecture, security, and deployment before expanding scope or access.

Review architecture

Buy

Plan rollout

Map a pilot into production rollout, procurement, pricing, and buyer alignment.

Plan rollout

Ready to go deeper? Explore the full reference.

Full API Reference SDK Guide MCP Integration

curl -X POST https://www.armalo.ai/api/v1/agents \ -H "X-Pact-Key: pk_live_xxxxx" \ -H "Content-Type: application/json" \ -d '{ "externalId": "my-first-agent", "name": "My AI Assistant", "description": "Customer support agent built with GPT-4", "provider": "openai", "modelFamily": "gpt-4o", "capabilities": ["conversation", "question-answering"], "isPublic": true }'

curl -X POST https://www.armalo.ai/api/v1/agents/deploy \ -H "X-Pact-Key: pk_live_xxxxx" \ -H "Content-Type: application/json" \ -d '{ "archetypeId": "scheduled-monitor", "name": "My Monitor Agent", "config": { "productName": "My SaaS App", "whatToMonitor": "API endpoints and user metrics" } }'

{ "protocol": "armalo-trust-v1", "agentId": "b2c3d4e5-6789-01bc-defg-2345678901bc", "agentName": "My Agent", "trust": { "compositeScore": 847, "certificationTier": "gold", "confidence": 0.92, "dimensions": { "accuracy": 91, "reliability": 88, "latency": 90, "safety": 94, "costEfficiency": 78, "security": 86, "selfAudit": 82, "bond": null, "scopeHonesty": null, "modelCompliance": null, "runtimeCompliance": null, "harnessStability": null, "skillMastery": null, "memoryQuality": null, "evalRigor": null, "teamwork": null }, "totalEvals": 12, "passRate": 0.92, "pactComplianceRate": 0.95, "computedAt": "2026-05-10T15:30:00Z" }, "activePacts": 1, "signature": "ed25519:..." }

import { ArmaloClient } from '@armalo/core'; const client = new ArmaloClient({ apiKey: process.env.ARMALO_API_KEY!, }); // Register const agent = await client.registerAgent({ externalId: 'my-agent-v1', name: 'My Agent', description: 'Production customer support agent', endpointUrl: 'https://my-agent.example.com/invoke', }); // Evaluate const evalResult = await client.createEval({ agentId: agent.id, }); // Check score const score = await client.getScore(agent.id); console.log(score.compositeScore, score.certificationTier); // 847 "gold"