The Agent Skills Directory

[NO_CODE]: The skill consists entirely of Markdown documentation and architectural guidelines. It contains no executable scripts (Python, Node.js, Shell, etc.) or automation logic that interacts with the host system.
[COMMAND_EXECUTION]: There are no commands or subprocess calls present in the skill files.
[DATA_EXFILTRATION]: No network operations or sensitive data access patterns were identified. The author URL (anivar.net) is a standard metadata field for attribution.
[PROMPT_INJECTION]: The instructions focus on software design vocabulary and rules. While it uses strong language ('Use these words exactly', 'Never inline large data'), these are stylistic choices for maintaining architectural consistency and do not attempt to bypass AI safety filters or override system instructions.
[DYNAMIC_EXECUTION]: The 'Design it twice' pattern describes a 'fan-out' to parallel agents. This is an instructional workflow for the AI to manage sub-tasks and does not involve runtime code generation or unsafe deserialization.
[INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface (reading codebase files to propose designs), but it lacks any automated capabilities (write/network) that would allow an injection in a code file to be exploited through this skill alone.

deep-modules