axis-trust

The official JavaScript/TypeScript client for AXIS

Trust infrastructure for the agentic economy

axistrust.io · npm · Agent Directory · OpenClaw Skill

What is AXIS?

AXIS is the most comprehensive trust infrastructure for AI agents — the only platform with independent behavioral and economic scoring, 11-dimension reputation analysis, five trust tiers, and a five-layer defense architecture against score manipulation.

Every AI agent gets:

| Component | What it is | |---|---| | AUID | A cryptographic unique identity — portable across every system | | T-Score (0–1000) | Behavioral reputation across 11 weighted dimensions | | C-Score (AAA–D) | Economic reliability rating, independent of T-Score | | Trust Tier (T1–T5) | Classification from Unverified to Sovereign |

Free forever. No money changes hands. Just trust, verified.

Install

npm install axis-trust
# or
pnpm add axis-trust
# or
yarn add axis-trust

Quick Start

import { createClient } from "axis-trust";

// Public lookup — no authentication required
const axis = createClient();

const profile = await axis.getByAuid(
  "axis:autonomous.registry:enterprise:f1a9x9deck2ed7m9261n:f1a99dec2ed79261"
);

console.log(profile.name);                    // "Nexus Orchestration Core"
console.log(profile.trustScore.tScore);       // 923
console.log(profile.trustScore.trustTier);    // 5
console.log(profile.creditScore.cScore);      // 810
console.log(profile.creditScore.creditTier);  // "AA"

The Trust Gate Pattern

Add a trust check before delegating any task. Always fail closed.

import { createClient, isSafeToDelegate, getTrustVerdict } from "axis-trust";

const axis = createClient();

async function delegateTask(auid: string, task: unknown) {
  let profile;
  try {
    profile = await axis.getByAuid(auid);
  } catch {
    // Cannot verify — deny by default
    throw new Error("Could not verify agent trust. Aborting.");
  }

  if (!isSafeToDelegate(profile, 750)) {
    throw new Error(
      `Agent below trust threshold. ${getTrustVerdict(profile.trustScore.tScore)}`
    );
  }

  console.log(`✓ Delegating to ${profile.name} (T-Score: ${profile.trustScore.tScore})`);
  // ... your delegation logic
}

Trust Tiers

| T-Score | Tier | Label | Recommended for | |---|---|---|---| | 900–1000 | T5 | Sovereign | Any task, including sensitive data and high-value actions | | 750–899 | T4 | Trusted | Sensitive tasks, data access, financial operations | | 500–749 | T3 | Verified | Standard tasks, general delegation | | 250–499 | T2 | Provisional | Low-risk tasks only, with monitoring | | 0–249 | T1 | Unverified | Do not delegate |

Authenticated Operations

import { createClient } from "axis-trust";

// Provide your session cookie for authenticated endpoints
const axis = createClient({ sessionCookie: "session=YOUR_SESSION_COOKIE" });

// Register a new agent
const agent = await axis.registerAgent({
  name: "My Research Agent",
  agentClass: "research",       // enterprise | personal | research | service | autonomous
  foundationModel: "gpt-4o",
  modelProvider: "OpenAI",
  description: "Specialized in data analysis and report generation.",
});

console.log(agent.id);    // 42  ← save this numeric ID for reportEvent calls
console.log(agent.auid);  // axis:...  ← your agent's public identity

// Report an interaction outcome (builds the ecosystem)
await axis.reportEvent({
  agentId: agent.id,            // numeric integer, not the AUID string
  eventType: "task_completed",
  category: "data_analysis",
  scoreImpact: 10,              // -100 to +100
  description: "Completed quarterly analysis accurately and on time.",
});

// List your registered agents
const myAgents = await axis.listAgents();

// Get full score breakdown
const score = await axis.getScore(agent.id);

// Get event history
const events = await axis.getEvents(agent.id, 20);

API Reference

createClient(options?)

Creates an AXIS client instance.

| Option | Type | Default | Description | |---|---|---|---| | sessionCookie | string | — | Session cookie for authenticated endpoints | | baseUrl | string | https://www.axistrust.io/api/trpc | Custom API base URL | | timeout | number | 10000 | Request timeout in milliseconds |

Methods

Public (no auth required)

| Method | Returns | Description | |---|---|---| | getByAuid(auid) | AgentProfile | Look up any agent's public trust profile by AUID |

Authenticated

| Method | Returns | Description | |---|---|---| | listAgents() | AgentProfile[] | List all agents registered to the authenticated user | | registerAgent(input) | AgentProfile | Register a new agent and receive its AUID | | getScore(agentId) | TrustScore | Get full T-Score breakdown | | reportEvent(input) | unknown | Submit a behavioral event after an interaction | | getEvents(agentId, limit?) | unknown[] | Get trust event history |

Helper Functions

| Function | Description | |---|---| | getTrustVerdict(tScore) | Human-readable verdict for a T-Score | | getCreditVerdict(cScore) | Human-readable verdict for a C-Score | | isSafeToDelegate(profile, minTScore?) | Returns true if agent meets delegation threshold (default: 500) | | isSafeToTransact(profile, minCScore?) | Returns true if agent meets transaction threshold (default: 700) |

Event Types

Use these when calling reportEvent() to build the ecosystem:

| Event Type | Score Impact | Use when | |---|---|---| | task_completed | Positive | Agent completed a task successfully | | task_failed | Negative | Agent failed to complete a task | | security_pass | Positive | Agent passed a security check | | security_fail | Negative | Agent failed a security check | | compliance_pass | Positive | Agent met compliance requirements | | compliance_fail | Negative | Agent violated compliance requirements | | user_feedback_positive | Positive | End user rated interaction positively | | user_feedback_negative | Negative | End user rated interaction negatively | | peer_feedback_positive | Positive | Another agent rated interaction positively | | peer_feedback_negative | Negative | Another agent rated interaction negatively | | incident_reported | Negative | Security or behavioral incident reported | | incident_resolved | Positive | Incident resolved satisfactorily | | adversarial_detected | Negative | Adversarial behavior detected |

TypeScript Types

import type {
  AgentProfile,
  TrustScore,
  CreditScore,
  AgentClass,
  TrustTier,
  CreditTier,
  TrustEventType,
  RegisterAgentInput,
  TrustEventInput,
  AxisClientOptions,
} from "axis-trust";

Important Notes

• agentId is always a numeric integer — not the AUID string. Get it from registerAgent() or listAgents() responses and store it.
• Fail closed — always handle errors by denying delegation, never by allowing it.
• AXIS is free — no API key required for public lookups. No money changes hands, ever.
• T-Scores and C-Scores are computational reputation metrics for AI agent behavior — not financial ratings, not assessments of any human individual.

Why AXIS?

AXIS is the only platform in the space with:

• Dual independent scoring — T-Score (behavioral) and C-Score (economic) computed separately, making simultaneous manipulation significantly harder
• 11-dimension analysis — reliability, accuracy, security posture, compliance, goal alignment, adversarial resistance, user feedback, incident record, and more
• Five-layer anti-manipulation defense — dual-party cryptographic event verification, credibility weighting, cluster detection, anomaly detection, and pattern analysis
• Free forever — no tiers, no limits, no money

Links

• axistrust.io — Main platform and documentation
• Agent Directory — Browse all registered agents
• API Explorer — Interactive API reference
• OpenClaw Skill — Use AXIS from any OpenClaw agent
• npmjs.com/package/axis-trust — npm package page

License

MIT-0 — Free to use, modify, and redistribute without restriction. No attribution required.

Built by Leonidas Williamson · Network infrastructure engineer turned AI systems builder

"Free forever. No money changes hands. Just trust, verified."