The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied architecture data to generate security definitions. The attack surface is minimal as the output is restricted to documentation. Evidence: (1) Ingestion points: User requests for threat models and architecture reviews in SKILL.md. (2) Boundary markers: Present in the form of mandatory templates in examples/security-definition-document.md and principle alignment in references/security-principles.md. (3) Capability inventory: Markdown and Mermaid generation only; no code execution, system modifications, or network writes identified across any files. (4) Sanitization: The skill enforces a self-consistency checklist in references/writing-quality-sds.md to validate agent findings against runtime realities and platform limitations.
[EXTERNAL_DOWNLOADS]: The skill references vendor-owned documentation hosted on contributing.bitwarden.com for architectural standards, principles, and vocabulary. These are informational links to official resources and do not facilitate remote code execution or automated downloads.
[COMMAND_EXECUTION]: All tasks described in the skill involve generating static text or diagrams. No instructions are present that attempt to invoke shell commands, manage system processes, or perform network operations. Internal references to ${CLAUDE_PLUGIN_ROOT} are used exclusively for locating local reference documents.

threat-modeling