Security audit
Security checks for vulnerabilities and agentic risk
This is a Markdown-only safety workflow for reviewing and installing OpenClaw skills, with explicit review, sandbox, and consent gates.
Before installing, note that this skill can lead an agent to run ClawHub inspection and install commands and to read OpenClaw configuration during a user-directed install review. It is designed to avoid automatic trust changes; only approve any allowBundled write when you understand the exact skill slug and backup/rollback path.
The file mixes English metadata with Chinese user-facing description and trigger phrases, but it does not indicate that the skill is intentionally limited to Chinese-speaking users or provide an opt-in language choice. This can violate a language/locale policy when users are forced into a specific language without explicit justification.
The manifest description says to use the skill when asked to 'install a skill safely,' 'inspect skill permissions,' 'review third-party skill risk,' or 'run a pre-install security audit,' but it does not define tighter trigger constraints or negative examples. In a markdown/manifest context, this broad phrasing could cause the skill to activate for general review or audit conversations beyond the intended install workflow.
The instruction 'Proceed only after explicit “yes/install/继续”' embeds a specific non-default locale term in the required consent language. This can be a language/locale policy issue because it imposes particular accepted phrasing without stating that users may respond in their preferred language.
No suspicious patterns detected.
