The Agent Skills Directory

[SAFE]: The skill serves as a comprehensive guide for designing AI agent architectures. It includes code snippets for various patterns such as ReAct (Reason + Act), Plan-and-Execute, and Reflexion, all of which are standard industry practices.
[COMMAND_EXECUTION]: The skill includes a 'companion check' in the SKILL.md file that performs a directory listing using ls to check for recommended skills. This is a benign discovery operation intended to assist the user with setup and does not pose a security risk.
[DATA_EXFILTRATION]: No evidence of data exfiltration was found. All network-related examples (like web search tool schemas) are standard architectural templates and do not contain hardcoded malicious endpoints.
[REMOTE_CODE_EXECUTION]: There are no patterns involving the download and execution of remote scripts. The installation instructions use npx skills add, which is the standard mechanism for adding skills to supported AI agents.
[PROMPT_INJECTION]: The skill does not contain any instructions that attempt to override the AI agent's safety filters or core instructions. It follows best practices by recommending the use of delimiters and validation for external data processing.

ai-agent-design