The Agent Skills Directory

[COMMAND_EXECUTION]: The skill enforces an advisory-only role by strictly instructing the agent not to write code, create files, or run commands. It only generates analysis and optimized prompt text.- [PROMPT_INJECTION]: The skill includes instructions to prevent users from bypassing its constraints, specifically advising the agent to refuse execution requests and maintain its role as a prompt consultant.- [SAFE]: The skill reads standard project configuration files (e.g., package.json, go.mod) to detect the tech stack for context. This is routine behavior for developer tools and does not involve accessing sensitive credentials.

prompt-optimizer