← 返回 Skills 市场
private computation
作者
Justin Liu
· GitHub ↗
· v1.0.0
· MIT-0
273
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install private-computation
功能描述
Zero-Knowledge Execution for Sensitive Agent Tasks - Privacy computing framework for AI Agents
安全使用建议
This registry entry is an instruction-only README that promises advanced privacy features but contains no code or install metadata. Before using it or running the suggested npm install: 1) Verify the package exists on npm (or the GitHub repo) and inspect its source code — do not blindly npm install. 2) Confirm where audit logs are stored or transmitted, and how the masterKey is generated and protected. 3) Avoid storing real production secrets (medical, banking, or live LLM keys) until you can audit the implementation. 4) If you need these features for compliance, prefer well-audited libraries or vendor-provided solutions and consider an independent security review. If you want, I can: (a) check npm or GitHub for the referenced package and report what I find, or (b) list specific questions to ask the package author to verify safety.
功能分析
Type: OpenClaw Skill
Name: private-computation
Version: 1.0.0
The skill bundle provides documentation and usage instructions for a privacy-focused computation framework ('openclaw-private-computation') designed for AI agents. The SKILL.md file outlines features for encrypted secret management (AES-256-GCM), secure task execution, and audit logging, which are consistent with its stated purpose of enhancing security and compliance (GDPR/HIPAA). No evidence of malicious intent, data exfiltration, or harmful prompt injection was found in the provided files (_meta.json, SKILL.md).
能力评估
Purpose & Capability
The SKILL.md advertises encryption, TEE isolation, ZK proofs and blockchain-style audit logs and shows npm install instructions, but the registry entry contains only the SKILL.md (no code files, no install spec). That makes the claimed capabilities unverifiable from this bundle and indicates a mismatch between the described functionality and what the skill actually provides.
Instruction Scope
Instructions show API usage patterns that store and retrieve secrets (agent.setSecret / getSecret), execute sensitive tasks, and write audit logs (storagePath defaults to ~/.openclaw). These calls are in-scope for a private-computation library, but the document gives no details about where audit logs are transmitted, how the masterKey is stored/derived, or how isolation is achieved — missing details that matter for secrets handling.
Install Mechanism
Although the README shows 'npm install openclaw-private-computation' and 'clawhub install', the registry provides no install spec and no packaged code. That difference is suspicious: either this is documentation-only (which is fine) or the intended package must be fetched from an external registry (which should be audited before use).
Credentials
The skill declares no required environment variables or credentials, yet the examples demonstrate storing/using many sensitive API keys (OPENAI_API_KEY, MEDICAL_API_KEY, BANK_API_KEY, etc.). Absence of declared required envs is not itself dangerous, but you should not assume the skill (or the separately published npm package) treats or transmits those secrets safely without review.
Persistence & Privilege
The skill does not request always:true and is user-invocable only. However, the documented default storagePath (~/.openclaw) implies the library (if installed) would write to the user's home directory — verify storage/encryption details before saving secrets.
如何使用
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install private-computation - 安装完成后,直接呼叫该 Skill 的名称或使用
/private-computation触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
Initial release of OpenClaw Private Computation.
- Introduces a privacy-first computation framework for AI Agents, supporting zero-knowledge execution of sensitive tasks.
- Features encrypted credential storage (AES-256-GCM), isolated/secure task execution, and immutable audit logs.
- Supports multiple security levels, including basic, TEE isolation, and strict zero-knowledge modes.
- Designed for GDPR and HIPAA compliance.
- Provides a simple TypeScript/Node.js API for credential management, secure task execution, and audit logging.
- Documentation, examples, and roadmap included.
元数据
常见问题
private computation 是什么?
Zero-Knowledge Execution for Sensitive Agent Tasks - Privacy computing framework for AI Agents. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 273 次。
如何安装 private computation?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install private-computation」即可一键安装,无需额外配置。
private computation 是免费的吗?
是的,private computation 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
private computation 支持哪些平台?
private computation 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 private computation?
由 Justin Liu(@zhenstaff)开发并维护,当前版本 v1.0.0。
推荐 Skills