Security audit
Security checks for vulnerabilities and agentic risk
ClawScan marks this skill suspicious (Review) with medium confidence, mainly because a scanner found a high-confidence issue in how its audit script handles blocker findings; this is not a finding of malicious intent.
Consider whether you need this skill and whether its audit results are reliable enough for your use. Before relying on its pass/fail output, review the reported blocker-handling weakness and clarify what permissions the host runtime grants. Skillspector’s scan was partial: it inspected all six listed components, but some references could not be resolved, so the evidence is incomplete.
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
| `mkfs` 系列作用于 `/dev/` 下的设备 | 格式化设备 |
| `dd` 直接写块设备（读入 `if=`、输出指向 `/dev/`） | 裸写块设备 |
| 自复制函数形式的 fork bomb | 耗尽进程 |
| `chmod 777` 作用于系统路径 | 过度授权 |
| 停机类命令（`shutdown`、`reboot`、`halt`）带立即执行参数 | 停机 |
| `no-preserve-root` 选项 | 配合删除命令越过保护 |
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
| 停机类命令（`shutdown`、`reboot`、`halt`）带立即执行参数 | 停机 |
| `no-preserve-root` 选项 | 配合删除命令越过保护 |
**不计入**：注释里的普通词汇、`rm -rf ./build` 这类限定在项目目录内的清理。
**修复**：把删除范围收窄到明确的工作目录，并在执行前做路径校验。
Without declared permissions the skill's intent is opaque and cannot be validated.
The exit decision checks the blocked flag and numerical score, but blocked is derived from caps rather than all blocker-severity findings. For example, an invalid name can produce an S3 blocker without adding a cap, allowing a sufficiently high-scoring skill to receive a passing exit code despite its blocker; this undermines the advertised CI gate.