The Agent Skills Directory

[SAFE]: The skill is purely instructional, providing templates and rules for the agent to follow when presenting architectural proposals and code diffs.
[PROMPT_INJECTION]: No attempts to override system prompts, bypass safety filters, or use adversarial role-play patterns were detected.
[DATA_EXFILTRATION]: No evidence of sensitive data access, credential harvesting, or unauthorized network communication. References to other local skill files (writing-voice/SKILL.md, progress-summary/SKILL.md) are standard for skill-to-skill referencing.
[REMOTE_CODE_EXECUTION]: There are no commands for downloading external scripts, installing unverified packages, or executing dynamic code from untrusted sources.
[COMMAND_EXECUTION]: The skill does not invoke shell commands or interact with the operating system in any privileged or dangerous manner.

change-proposal