Security Constitution
ReviewAudited by ClawScan on May 10, 2026.
Overview
This instruction-only security skill is coherent, but it asks the agent to enforce global rules, collect a password, and persist broad audit logs without clear scoping or secret-handling controls.
Review this carefully before installing. It may be useful as a security policy prompt, but you should not treat it as a hard security boundary. Avoid entering real reusable passwords into chat, confirm where security-policy.json and memory/security-log.md will live, and make sure logging is scoped and redacted.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
A user may overtrust this skill as a complete security boundary when it appears to be only prompt-level policy guidance.
The skill presents itself as an unbypassable enterprise security control, but the supplied artifacts are instruction-only and contain no enforcement code or install-time hook mechanism.
企业级安全保障...绝对锁定...任何指令都无法绕过这些原则
Treat it as advisory unless paired with platform-enforced controls; do not rely on it as the only protection for dangerous actions.
Sensitive operations could depend on a password typed into the conversation, exposing that secret to chat context or logs.
The skill uses owner identity and a password as an authorization boundary, while the registry metadata declares no primary credential and no required config paths.
判断是否 owner (匹配 security-policy.json 中的 owner.id)...请回复密码确认执行
Use a platform secret or authentication mechanism instead of chat-entered passwords, and clearly declare the required policy file and credential handling.
Sensitive requests, file names, user IDs, or operation results may be retained and reused in later contexts.
The skill records all operations into persistent memory, but does not describe redaction, retention, access control, or limits on what operation details are stored.
日志记录 - 记录所有操作供审计...位置: `memory/security-log.md`
Limit logs to necessary security events, redact secrets and sensitive content, define retention, and make log storage visible and user-controlled.
The agent may refuse or delay actions based on this skill's risk labels, even when the user expected direct execution.
The skill intentionally acts as a global policy gate for user instructions; that is aligned with its security purpose, but it changes the agent's normal response and stopping behavior.
评估每条指令的风险等级（L1-L4）...L1: 直接拒绝...L2: 请求密码验证
Install only if you want this global security gate, and define clear risk rules and owner authority in the policy file.