Security audit
Security checks for vulnerabilities and agentic risk
This skill is a coherent UIBC bio-domain judging aid with a local criteria checker and no evidence of hidden access, exfiltration, persistence, or unsafe execution.
Installers should treat this as a specialized UIBC/BIO-domain judging skill and invoke it only when the user explicitly wants TH-BIO-001 or bio-domain accountability analysis; ambiguous mentions of generic terms like memory transfer or responsibility should be clarified before using it.
The skill advertises a long list of generic trigger terms such as “UIBC”, “生物基因”, “记忆迁移”, and related governance phrases without clear activation boundaries, exclusions, or disambiguation rules. This can cause the skill to activate in unrelated contexts, increasing the chance that an agent follows domain-specific instructions or outputs inappropriately when a user merely mentions overlapping terms.
The skill declares broad natural-language trigger conditions in metadata and description, including generic terms like UIBC, 生物基因, 迁移判定, 证据包, and 责任归属. This can cause the agent to invoke the skill outside narrowly intended BIO-domain judging scenarios, leading to inappropriate authority framing, irrelevant policy injection, or misrouting user requests into this skill's rigid adjudication workflow.
The '何时用' section repeats broad and ambiguous trigger wording, especially around cross-domain review and common terms like 记忆迁移 or 责任归属, without clear scope boundaries. In practice, this increases the chance that unrelated conversations are pulled into a specialized evidentiary and accountability framework, which can distort responses and create unsafe over-application of this skill's domain assumptions.
