Security audit
Security checks for vulnerabilities and agentic risk
This read-only skill helps evaluate agent token-cost changes and does not show hidden, destructive, or exfiltrating behavior.
Install this only if you are comfortable letting the agent review token, session, cron, and before/after operational details for the agents you name. Redact secrets and private log contents before sharing evidence, and treat any rollback or configuration change as a separate human-reviewed action.
The file lists a single activation phrase, "eval agent cost change," but does not clarify the context in which it should activate, any constraints on its use, or examples of when it should not trigger. In a markdown file, this can qualify as a vague trigger because the invocation scope is not explicitly bounded.
This markdown file contains mixed-language instructional content on L79 (English plus Chinese text) without any note that multilingual output is optional or intended. That can conflict with language/locale policy expectations because the skill examples may implicitly steer responses into a specific language without user opt-in.
