disable-model-invocation

Manual model invocation requires setting the frontmatter flag disable-model-invocation: true. This enforces human oversight by blocking models from triggering automated actions in sensitive AI workflows.

Manual invocation in AI safety is a guardrail that prevents automated model execution by requiring explicit human activation. It enforces human oversight to protect sensitive actions and testing.

You need to disable model invocation for security-conscious workflows requiring strict human oversight. This guardrail prevents accidental or unauthorized model-triggered actions during sensitive tasks and testing.

To configure a skill for manual activation only, apply the frontmatter flag disable-model-invocation: true to mark manual invocability. This prevents models from initiating the skill automatically.

No, this approach strictly blocks model-initiated execution to enforce manual activation for sensitive testing. It serves as a guardrail ensuring actions require explicit human commands rather than automated model triggers.