Home/Integrations/ZeroClaw
Z
+TORK

ZeroClaw + Tork Network

Zero-Trust Security for AI Agents

Sandboxing isolates execution. Tork adds what sandboxing can't — PII detection, compliance attestation, and verifiable trust.

Sandbox
Isolation
+ PII
Detection
+ Receipts
Compliance
+ Badges
Trust
Quick Start

Add Tork in 3 Steps

1

Install the Tork SDK

Add Tork governance on top of ZeroClaw's sandbox.

npm install @torknetwork/node-sdk
2

Layer Governance Over Sandbox

Tork wraps the sandbox boundary — scan data going in and coming out.

import { TorkClient } from '@torknetwork/node-sdk';

const tork = new TorkClient({
  apiKey: process.env.TORK_API_KEY,
  config: {
    redactPii: true,
    generateReceipts: true,
  },
});
3

Verify End-to-End

Confirm sandbox isolation + Tork governance work together.

// Sandbox blocks execution risks
// Tork blocks data risks (PII, compliance)
const receipt = await tork.govern('test string');
console.log('Governance active:', receipt.receiptId);
Integration Code

Full Integration Example

import { ZeroClaw, Sandbox } from 'zeroclaw';
import { TorkClient } from '@torknetwork/node-sdk';

const tork = new TorkClient({
  apiKey: process.env.TORK_API_KEY,
  config: {
    redactPii: true,
    generateReceipts: true,
  },
});

// Create a sandboxed agent with governance
const agent = new ZeroClaw({
  name: 'secure-agent',
  sandbox: new Sandbox({
    network: 'restricted',
    filesystem: 'read-only',
    processes: 'blocked',
  }),
});

// Tork governance layer — scans data at the sandbox boundary
agent.onSandboxInput(async (data) => {
  // Scan data entering the sandbox for PII
  const governed = await tork.govern(data.content);

  if (governed.piiDetected.length > 0) {
    console.log('PII caught at sandbox boundary:', governed.piiDetected);
    data.content = governed.governedText;
  }

  return data;
});

agent.onSandboxOutput(async (data) => {
  // Scan data leaving the sandbox
  const governed = await tork.govern(data.content);

  return {
    content: governed.governedText,
    compliance: {
      receiptId: governed.receiptId,
      piiRedacted: governed.piiDetected,
      sandboxed: true,
      governed: true,
    },
  };
});

// Security stack: ZeroClaw sandbox + Tork governance
// - Sandbox: prevents code execution, network abuse, file access
// - Tork: prevents PII leaks, provides compliance proof, trust badges
await agent.start();
Comparison

Without Tork vs. With Tork

Without Tork
With Tork
Sandbox blocks execution but not data leaks
Sandbox + PII detection = complete protection
No compliance proof from isolation alone
Immutable receipts for every governed interaction
Sandboxed agents have no trust signal
Trust badges prove sandbox + governance compliance
No visibility into what data passes through
50+ PII types detected at sandbox boundary
Isolation doesn't equal compliance
TORKING-X score proves governance quality
No audit trail for sandboxed operations
Full audit log with receipt chain
Trust Badge

Show Users Your Agent is Governed

Embed a verifiable trust badge to prove your agent passes Tork governance checks.

<!-- ZeroClaw + Tork Trust Badge -->
<a href="https://tork.network/badge/YOUR_BADGE_ID">
  <img
    src="https://tork.network/api/v1/badge/svg/YOUR_BADGE_ID"
    alt="Tork Governed — ZeroClaw Sandbox"
  />
</a>

Ready to Govern Your ZeroClaw Agents?

Add independent governance in under 5 minutes. Free tier available.

Get Started FreeView Documentation