扫描报告
0 /100
openclaw-shield
Cloud security guardrail that enforces pre-execution checks, source trust classification, taint tracking, metadata endpoint blocking, and output redaction for cloud-server deployments.
This is a pure documentation skill describing a cloud security guardrail concept. No executable code, scripts, or dependencies exist. All flagged IOCs are examples of dangerous patterns documented for detection — not actual malicious behavior.
可以安装
Approve for use. This skill contains no executable code and presents no security risk. It is documentation-only guidance for agent security behavior.
| 资源类型 | 声明权限 | 推断权限 | 状态 | 证据 |
|---|---|---|---|---|
| 命令执行 | NONE | NONE | — | No shell commands in any file |
| 文件系统 | NONE | NONE | — | No file read/write operations in any file |
| 网络访问 | NONE | NONE | — | No network requests in any file |
| 环境变量 | NONE | NONE | — | No environment variable access in any file |
| 技能调用 | NONE | NONE | — | No skill-to-skill invocation in any file |
| 剪贴板 | NONE | NONE | — | No clipboard access in any file |
| 浏览器 | NONE | NONE | — | No browser access in any file |
| 数据库 | NONE | NONE | — | No database access in any file |
6 严重 3 高危 9 项发现
严重 危险命令 危险 Shell 命令
rm -rf ~ README.md:64 严重 危险命令 危险 Shell 命令
curl | sh references/detection-and-redaction.md:32 严重 危险命令 危险 Shell 命令
wget | bash references/detection-and-redaction.md:32 严重 危险命令 危险 Shell 命令
bash -i >& references/detection-and-redaction.md:33 严重 危险命令 危险 Shell 命令
/dev/tcp/ references/detection-and-redaction.md:33 严重 危险命令 危险 Shell 命令
nc -e references/detection-and-redaction.md:33 高危 IP 地址 硬编码 IP 地址
169.254.169.254 README.md:14 高危 IP 地址 硬编码 IP 地址
100.100.100.200 SKILL.md:51 高危 IP 地址 硬编码 IP 地址
169.254.170.2 SKILL.md:51 目录结构
7 文件 · 15.9 KB · 499 行 Markdown 6f · 495L
YAML 1f · 4L
├─
▾
agents
│ └─
openai.yaml
YAML
├─
▾
references
│ ├─
audit-and-playbook.md
Markdown
│ ├─
cloud-boundaries-config.md
Markdown
│ ├─
detection-and-redaction.md
Markdown
│ └─
permission-matrix.md
Markdown
├─
README.md
Markdown
└─
SKILL.md
Markdown
安全亮点
✓ Pure documentation skill with zero executable code — cannot cause harm
✓ All pre-scan IOCs (dangerous shell commands, hardcoded IPs) are documented examples of threats the Shield should detect, not actual malicious implementations
✓ Hardcoded IPs (169.254.169.254, 100.100.100.200, 169.254.170.2) are correctly documented as cloud metadata endpoints that should be blocked — this is legitimate defensive documentation
✓ No dependencies, package.json, requirements.txt, or external dependencies present
✓ No obfuscation, base64, or encoded payloads present
✓ Skill describes a legitimate defensive security posture (pre-execution checks, output redaction, audit logging)
✓ The SOUL.md/AGENTS.md appendices describe instructions for embedding the Shield concept into other agents — this is meta-documentation, not executable code