The Agent Skills Directory

[SAFE]: The skill is primarily composed of architectural documentation, design patterns, and cross-platform references for building plugin systems. It promotes security best practices such as declarative manifests and host-owned trust boundaries.
[EXTERNAL_DOWNLOADS]: The skill contains numerous links to official documentation and repositories from well-known services and trusted organizations, including Anthropic (claude.com), GitHub (github.com), and OpenAI (openai.com). These references are used for informational purposes and do not trigger untrusted downloads or remote code execution.
[COMMAND_EXECUTION]: The provided Python script (scripts/validate_manifest.py) is a static validation tool that uses standard libraries (json, re, argparse). It performs local file reads and regex checks to ensure manifest consistency without executing dangerous shell commands or performing network operations.
[PROMPT_INJECTION]: No malicious prompt injection vectors were detected. The skill contains documentation about the !command syntax used by other platforms, but does not use it to execute hidden commands. The instructions explicitly guide users toward maintaining strict safety filters and policy controls.

ai-coding-agents-plugins