The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's configuration includes pre and post lifecycle hooks that execute npm test and memory_store commands. These are standard operations for a development-focused skill intended to verify code changes and manage internal state.
[DATA_EXPOSURE]: The provided code examples (e.g., Authentication Service, Performance Testing) use mock repositories, local variables, and generic placeholders like user@example.com. There are no hardcoded secrets or patterns indicating sensitive data exposure.
[PROMPT_INJECTION]: The instructions are clearly focused on defining the agent's role as a refinement specialist. There are no attempts to bypass safety filters, override system instructions, or extract system prompts.
[REMOTE_CODE_EXECUTION]: The skill does not download external scripts or execute code from remote or untrusted sources. All logic and examples are contained within the skill definition.

agent-refinement