by 安全扫描
OpenClaw
可疑
medium confidenceThe skill's files and instructions broadly match an installer for an open-source security agent, but several inconsistencies and high-impact options (auto-kill, quarantine, cloud API, AI correlation) and metadata mismatches warrant caution before installing.
评估建议
This package appears to be a legitimate installer for an on-host security agent, but review these points before installing:
- Verify provenance: compare the repository and release referenced in SKILL.md to the official project (the skill points to github.com/1sec-security/1sec). Confirm the GitHub owner is who you expect and inspect the release artifacts and checksums yourself. The registry's reported version (0.4.15) differs from the files (0.4.11) — resolve that mismatch.
- Prefer manual ins...详细分析 ▾
ℹ 用途与能力
The skill claims to install and manage the 1-SEC security binary and the included script and docs do exactly that (download a release binary, verify checksum, install, run setup, apply presets). That capability is proportional to the stated purpose. However: the registry metadata (version 0.4.15) does not match the SKILL.md and installer script (version 0.4.11), and the package provides no homepage while pointing to an external GitHub repo in SKILL.md. These mismatches reduce confidence in provenance.
ℹ 指令范围
Runtime instructions and the installer script are narrowly scoped to downloading a specific GitHub release binary, verifying its SHA256, installing it, running non-interactive setup, and applying a preset (default: safe/dry-run). The README/runbooks include high-impact operational instructions (vps-agent preset with skip_approval: true that will kill processes and quarantine files, cloud API options, webhook notifications). Those powerful enforcement behaviors are documented and opt-in, but they are present in the instructions and can be enabled by the operator — users should be aware of the destructive actions the product can take if presets are changed.
✓ 安装机制
No opaque third-party downloads or pastebin URLs are used: the installer script fetches binaries and checksums directly from GitHub Releases and verifies SHA256 before installing. The script explicitly avoids piping remote content to shell, deletes mismatched downloads, and documents manual verification. This is an expected and reasonably safe install approach if you trust the GitHub release source and account.
ℹ 凭证需求
No environment variables are required by the skill, and SKILL.md only lists optional env vars (GEMINI_API_KEY, ONESEC_API_KEY, ONESEC_WEBHOOK_URL). The references mention additional env patterns (e.g., GEMINI_API_KEY_2) and cloud API keys for centralized management; these are optional but carry privacy/credential implications. Enabling cloud or AI correlation will transmit alert metadata/host identifiers off-host (documented as opt-in). Requesting webhook URLs and API keys is proportional to the product's features but merits operator review before enabling.
ℹ 持久化与权限
The skill does not request 'always: true' and does not force persistent installation by itself; it installs a binary that is intended to run as a local daemon. The product recommends running with sudo/root for full enforcement, which is consistent with a host security agent but increases privilege impact. The runbooks document auto-update/selfupdate behavior (checks for updates daily) and an opt-in cloud control plane — both normal for agents but notable from an operational-security perspective.
安全有层次，运行前请审查代码。
运行时依赖
无特殊依赖
安装命令
点击复制官方npx clawhub@latest install 1sec-security
镜像加速npx clawhub@latest install 1sec-security --registry https://cn.longxiaskill.com 镜像可用
本土化适配说明
1-SEC: All-in-One Cybersecurity for AI Agent Hosts — 安全防护工具 安装说明： 安装命令：["openclaw skills install 1sec-security","npx clawhub@latest install 1sec-security"] 支持国内镜像加速，使用 --registry https://cn.longxiaskill.com 参数可加速下载