private-computation（私人计算）

OpenClaw Private Computation

Zero-Knowledge Execution for Sensitive Agent Tasks AI Agent 敏感任务的零知识执行框架

A privacy-first computation framework for AI Agents that need to process sensitive data securely. Built with TypeScript for the Node.js ecosystem.

为需要安全处理敏感数据的 AI Agent 打造的隐私优先计算框架。使用 TypeScript 为 Node.js 生态系统构建。

✨ Features / 功能特性

• 🔐 Encrypted Credential Storage / 加密凭证存储

- AES-256-GCM encryption for API keys and secrets - API 密钥和敏感信息的 AES-256-GCM 加密

• 🛡️ Secure Task Execution / 安全任务执行

- Isolated execution environment for sensitive operations - 敏感操作的隔离执行环境

• 📝 Immutable Audit Trail / 不可篡改审计日志

- Blockchain-style audit logs for compliance - 区块链式审计日志，满足合规要求

• 🎯 Multiple Security Levels / 多层安全级别

- Basic, Standard (TEE), and Strict (Zero-Knowledge) - 基础、标准（TEE）、严格（零知识证明）

• ✅ GDPR & HIPAA Compliant / GDPR 和 HIPAA 合规

- Designed for regulatory compliance - 专为满足监管合规要求而设计

📦 Installation / 安装

Via npm

npm install openclaw-private-computation

Via ClawHub

clawhub install private-computation

🚀 Quick Start / 快速开始

import { PrivateAgent } from 'openclaw-private-computation';

// Initialize agent / 初始化 Agent
const agent = new PrivateAgent({
  securityLevel: 'basic',  // basic | standard | strict
  audit: true              // 启用审计日志
});

// Store credentials securely / 安全存储凭证
await agent.setSecret('OPENAI_API_KEY', 'sk-...');

// Execute sensitive tasks / 执行敏感任务
const result = await agent.executeTask(async () => {
  const apiKey = await agent.getSecret('OPENAI_API_KEY');
  return await callAPI(apiKey);
}, {
  audit: true,
  metadata: { taskType: 'api_call' }
});

console.log(result);
// { status: 'success', data: {...}, auditId: '...', executionTime: 123 }

🎯 Use Cases / 使用场景

1. Medical AI (HIPAA) / 医疗 AI（HIPAA 合规）

const diagnosis = await agent.executeTask(async () => {
  const key = await agent.getSecret('MEDICAL_API_KEY');
  return await analyzeMedicalData(patientData, key);
}, {
  audit: true,
  metadata: { complianceLevel: 'HIPAA' }
});

2. Financial Services (PCI-DSS) / 金融服务（PCI-DSS 合规）

const transaction = await agent.executeTask(async () => {
  const bankKey = await agent.getSecret('BANK_API_KEY');
  return await processPayment(amount, bankKey);
}, {
  audit: true,
  timeout: 30000
});

3. AI Agent with Private Context / 带私有上下文的 AI Agent

const response = await agent.executeTask(async () => {
  const llmKey = await agent.getSecret('LLM_API_KEY');
  // Private context never exposed / 私有上下文永不暴露
  return await generateAI(userQuery, privateContext, llmKey);
}, {
  audit: true
});

🔧 API Reference / API 文档

Configuration / 配置

const agent = new PrivateAgent({
  securityLevel: 'basic' | 'standard' | 'strict',
  encryption: 'aes-256-gcm' | 'chacha20-poly1305',
  audit: boolean,
  storagePath: string,  // Default: ~/.openclaw
  masterKey: string     // Auto-generated if not provided / 未提供时自动生成
});

Security Levels / 安全级别

Credential Management / 凭证管理

// Store a secret / 存储密钥
await agent.setSecret('KEY_NAME', 'secret-value');

// Retrieve a secret / 获取密钥
const value = await agent.getSecret('KEY_NAME');

// List all secrets / 列出所有密钥（仅键名）
const keys = agent.listSecrets();

// Delete a secret / 删除密钥
await agent.deleteSecret('KEY_NAME');

Task Execution / 任务执行

const result = await agent.executeTask(async () => {
  // Your task logic / 你的任务逻辑
  return someData;
}, {
  audit: true,           // Enable audit logging / 启用审计日志
  proof: true,           // Generate ZK proof / 生成零知识证明 (strict mode)
  timeout: 30000,        // Timeout in ms / 超时时间（毫秒）
  metadata: {...}        // Custom metadata / 自定义元数据
});

Audit & Compliance / 审计与合规

// Get audit logs / 获取审计日志
const logs = agent.getAuditLogs(10);

// Verify integrity / 验证完整性
const integrity = agent.verifyAuditIntegrity();
// { valid: true }

// Get statistics / 获取统计信息
const stats = agent.getAuditStatistics();
// { totalLogs: 42, byAction: {...}, byResult: {...} }

🏗️ Architecture / 架构

OpenClaw Private Computation
│
├── Core Layer / 核心层
│   ├── Encryption Manager / 加密管理器 (AES-256-GCM)
│   ├── Credential Vault / 凭证保险库
│   ├── Audit Logger / 审计日志器 (Blockchain-style)
│   └── Task Executor / 任务执行器
│
├── Crypto Layer (Future) / 加密层（未来）
│   ├── zk-SNARKs (Zero-Knowledge Proofs) / 零知识证明
│   ├── TEE (Trusted Execution Environment) / 可信执行环境
│   └── Homomorphic Encryption / 同态加密
│
└── Integration Layer (Future) / 集成层（未来）
    ├── LangChain Adapter
    ├── Vercel AI SDK Adapter
    └── Claude API Adapter

🛣️ Roadmap / 路线图

Phase 1: MVP ✅ (Current / 当前)

• [x] Encrypted credential storage / 加密凭证存储
• [x] Basic audit logging / 基础审计日志
• [x] Simple SDK API / 简洁的 SDK API
• [x] Examples and documentation / 示例和文档

Phase 2: Zero-Knowledge (Next / 下一步)

• [ ] zk-SNARKs integration / zk-SNARKs 集成
• [ ] Proof generation and verification / 证明生成和验证
• [ ] Circuit library / 电路库

Phase 3: TEE Integration / TEE 集成

• [ ] Intel SGX support / Intel SGX 支持
• [ ] ARM TrustZone support / ARM TrustZone 支持
• [ ] Hybrid mode (TEE + zk) / 混合模式（TEE + 零知识）

Phase 4: Enterprise Features / 企业功能

• [ ] AI framework integrations / AI 框架集成
• [ ] Compliance reporting / 合规报告
• [ ] Multi-party computation / 多方计算

🔬 Technology / 技术栈

• Language / 语言: TypeScript
• Runtime / 运行时: Node.js 18+
• Encryption / 加密: Node.js Crypto (AES-256-GCM, ChaCha20-Poly1305)
• Zero-Knowledge / 零知识: snarkjs (coming soon / 即将推出)
• TEE: Intel SGX (planned / 规划中)
• Audit / 审计: Blockchain-inspired immutable logs / 区块链式不可篡改日志

📚 Documentation / 文档

• GitHub Repository / 代码仓库
• Quick Start Guide / 快速开始指南
• Full Documentation / 完整文档
• Examples / 示例

🌟 Why Choose OpenClaw? / 为什么选择 OpenClaw？

"The first production-ready privacy computing framework for TypeScript" "首个生产级 TypeScript 隐私计算框架"

No other TypeScript/JavaScript library provides: 其他 TypeScript/JavaScript 库都不提供：

• ✅ Zero-knowledge execution for AI agents / AI Agent 的零知识执行
• ✅ HIPAA/GDPR-ready audit trails / HIPAA/GDPR 就绪的审计追踪
• ✅ Simple API for complex cryptography / 复杂加密的简洁 API
• ✅ Open source and extensible / 开源且可扩展

🤝 Contributing / 贡献

We welcome contributions! Please see our GitHub repository for details.

欢迎贡献！详情请查看我们的 GitHub 仓库。

📄 License / 许可证

MIT-0 License - Free and open source, no attribution required. MIT-0 许可证 - 免费开源，无需署名。

Built for the AI Agent era. Secure by default. Private by design. 为 AI Agent 时代而生。默认安全。隐私优先。 🚀