by 安全扫描
OpenClaw
安全
high confidenceThis is an instruction-only governance guidance skill whose requested resources and examples match its stated purpose and do not ask for unrelated credentials or installs.
评估建议
This skill is documentation and example code for agent governance and appears internally consistent. Before copying examples into production: 1) restrict file paths when loading policies and avoid reading arbitrary system files; 2) sanitize and redact data before writing audit logs or sending them externally; 3) validate and test intent-detection regexes to avoid false positives/negatives and DoS via pathological regexes; 4) do not run example code unmodified — add input validation, error handli...详细分析 ▾
✓ 用途与能力
Name/description describe governance patterns and the SKILL.md contains architecture, policy code, intent-detection examples, YAML policy loading, and auditing guidance — all directly relevant to building agent governance. The skill declares no binaries, env vars, or installs, which is proportionate for a documentation/instruction skill.
ℹ 指令范围
The SKILL.md includes runnable code examples that open files (load_policy(path)), parse YAML, check content, and log audits. Those examples are appropriate for the stated purpose but do imply that an implementation will read policy/config files and produce logs. Consumers should avoid blindly executing example code or using overly-broad file paths; ensure policy-loading is restricted to trusted config paths and audit logs are redacted before storage or transmission.
✓ 安装机制
No install spec and no code files — this is instruction-only. No network downloads or package installs are requested.
✓ 凭证需求
The skill does not request environment variables, credentials, or config paths. Code snippets reference pattern-matching for secrets (e.g., api_key/secret/password) but do not attempt to access environment secrets themselves. This is proportionate for a governance guidance skill.
✓ 持久化与权限
No always:true, no installation, and no claimed modifications to other skills or global agent settings. Skill is user-invocable and can be called by the agent (normal behavior) but does not request persistent privileges.
安全有层次，运行前请审查代码。
运行时依赖
无特殊依赖
版本
latestv1.0.12026/3/9
Import from LeoYeAI/openclaw-master-skills on 2026-03-09
● 无害
安装命令
点击复制官方npx clawhub@latest install agent-governance
镜像加速npx clawhub@latest install agent-governance --registry https://cn.longxiaskill.com