Skip to main content

In this article

security-planning

security-planning
FieldValue
Kindskill
Source.github/skills/project-planning/security-planning
InvocationInvoked directly as /security-planning, or loaded on demand by referencing agents
InteractiveNo

What it does​

Security planning and plan-drift analysis for STRIDE, standards, controls, backlog handoff, current findings, and TM7 generation.

When to use it​

Use this skill directly or through Security Planner, Security Reviewer, or Code Review for STRIDE planning, control mapping, security backlog guidance, or correlation of a security baseline with supplied current findings. Drift analysis does not scan, verify findings, re-rate severity, or modify its inputs.

For threat-model files, a separate generation path produces TM7 and Markdown from one YAML/JSON specification. Native visual feedback is opt-in and requires Windows, an interactive desktop, UI Automation, and the pinned Threat Modeling Tool version. All outputs require qualified security review, not merely successful generation.

Example usage​

Ask: /security-planning Compare the attached sample security baseline with these current diff findings. Keep both inputs unchanged and recommend follow-up only. Supply finding locations, evidence scope, and the baseline's controls. Expect a correlation that distinguishes supported drift or new threats from conclusions suppressed for insufficient evidence. Default planning/customization exclusions remain visible; a narrow diff cannot establish repository-wide control validation.

For generation instead, provide a sanitized threat-model specification and request pre-populated-comprehensive or diagram-only-defer-to-tmt output. Success means the TM7 model and Markdown report derive from the same specification, with actual validation evidence and limitations reported separately.

Do not infer consent for the native feedback loop from a generation request: it takes control of mouse and keyboard until release. Feedback overlays remain pending human approval, and automated layout gates do not establish semantic security approval or permission to overwrite the canonical baseline.