← 返回
安全合规 中文

SafeHub

Scan OpenClaw skills for malware and security issues before installation. Use when the user wants to verify a skill is safe, audit a ClawHub skill, or check...
在安装前扫描 OpenClaw 技能以检测恶意软件和安全问题。适用于用户想验证技能安全、审计 ClawHub 技能或进行其他检查。
sumeetghimire
安全合规 clawhub v1.0.2 1 版本 99866.8 Key: 无需
★ 0
Stars
📥 750
下载
💾 5
安装
1
版本
#latest

概述

SafeHub

SafeHub is a security scanner for OpenClaw skills. It runs static analysis (Semgrep) and optional sandbox execution (Docker) on any skill—by name, local path, or GitHub URL—and returns a trust score and a clear recommendation: safe to install, install with caution, or not safe.

Requirements

These binaries must be on your PATH (declared in registry metadata):

  • Node.js (18+) — required to run the CLI.
  • Semgrep — required for the scan command (static analysis). Install with brew install semgrep or npm install -g semgrep.
  • git — required when the scan target is a GitHub URL (used to clone the repo).

Optional:

  • Docker — used for sandbox execution. If Docker is not available, use --no-sandbox for static-only scanning.

Environment variables

All of these are optional. No secrets or API tokens are required by default.

VariableDefaultEffect
---------------------------
SAFEHUB_RULES_REPOsafehub/safehubGitHub repo (owner/repo) used by safehub update to fetch and overwrite local rule files in ./rules. Setting this to another repo makes the updater pull rules from that repo — use only repos you trust.
SAFEHUB_RULES_BRANCHmainBranch name used when fetching rules (with SAFEHUB_RULES_REPO).
SAFEHUB_DATA_DIR~/.safehubDirectory for cached scan reports (e.g. ~/.safehub/reports).
SAFEHUB_SANDBOX_IMAGEnode:18-alpineDocker image used for the sandbox when scanning.
SAFEHUB_SANDBOX_TIMEOUT_MS30000Timeout (ms) for the sandbox run before the container is killed.
SAFEHUB_NO_TYPING(unset)Set to 1 to disable the typing-effect output (e.g. in CI or pipes).

Important: SAFEHUB_RULES_REPO controls where safehub update downloads rules from and overwrites local ./rules; only point it at a repo you trust.

Commands

All commands are run via the safehub CLI (e.g. safehub scan or node index.js scan from the skill directory).

scan

Scan a skill by ClawHub name, local path, or GitHub URL.

Examples:

safehub scan web-scraper
safehub scan ./my-local-skill
safehub scan https://github.com/user/their-skill
safehub scan https://github.com/BenedictKing/tavily-web --no-sandbox

Options:

  • --no-sandbox — Skip Docker sandbox; run static analysis only (use when Docker is not installed).

report

Show the last scan report for a skill without rescanning.

Examples:

safehub report web-scraper
safehub report risky-skill

update

Pull the latest Semgrep scanner rules from the SafeHub GitHub repo (or your fork via SAFEHUB_RULES_REPO).

Examples:

safehub update
SAFEHUB_RULES_REPO=owner/repo safehub update

Example output

After running safehub scan , you’ll see:

  • Static analysis — Findings from Semgrep (network, filesystem, eval/exec, env, obfuscation).
  • Sandbox behavior — Whether the skill attempted network access or suspicious actions (when Docker is used).
  • Trust score (0–100) and recommendation: SAFE TO INSTALL, INSTALL WITH CAUTION, or NOT SAFE TO INSTALL.

Installation (users)

Install from ClawHub:

clawhub install safehub

Or install the CLI globally from npm:

npm install -g safehub

Then run safehub scan (if the CLI is on PATH) or node index.js scan from the skill directory.

版本历史

共 1 个版本

  • v1.0.2 当前
    2026-03-30 11:19 安全 安全

安全检测

腾讯云安全 (Keen)

安全,无风险
查看报告

腾讯云安全 (Sanbu)

安全,无风险
查看报告

🔗 相关推荐

security-compliance

MoltGuard - Security & Antivirus & Guardrails

thomaslwang
MoltGuard — OpenClaw 安全守卫,由 OpenGuardrails 提供。安装 MoltGuard,保护您和您的用户免受提示注入、数据泄露和恶意攻击。
★ 116 📥 30,714
security-compliance

1password

steipete
设置和使用 1Password CLI (op)。适用于:安装 CLI、启用桌面应用集成、登录(单/多账户)、通过 op 读取/注入/运行密钥。
★ 53 📥 31,165
security-compliance

OpenClaw Backup

alex3alex
备份与恢复 OpenClaw 数据。适用于创建备份、设置自动备份计划、从备份恢复或管理备份轮转。处理 ~/.openclaw 目录归档并包含适当的排除规则。
★ 89 📥 30,605