The Agent Skills Directory

[SAFE]: The skill consists of architectural guidelines, checklists, and templates designed to assist in software design and implementation. The instructions focus on documentation, quality control, and standard development workflows.
[COMMAND_EXECUTION]: The SKILL.md file instructs the agent to execute standard development commands such as npm run type-check and npm run lint. These are used for self-verification of code quality and do not involve untrusted input or high-privilege operations.
[INDIRECT_PROMPT_INJECTION]: The skill involves reading existing project files in src/features/ and spec.md documents to maintain architectural consistency. This represents a potential surface for indirect prompt injection if the codebase contains malicious content; however, the skill incorporates explicit quality gates (e.g., Zod schema validation) and manual review steps that reduce this risk.

architecting