The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to utilize the omc CLI tool for managing the plugin lifecycle, involving the execution of commands like omc create, omc test, and omc add.
[REMOTE_CODE_EXECUTION]: The skill is designed to generate and execute arbitrary JavaScript, Python, and Bash scripts. These scripts are saved to the user's local filesystem and integrated into the statusline environment for execution.
[COMMAND_EXECUTION]: The PATTERNS.md file includes examples of using cachedExec to run shell commands within plugins. This creates a surface for command injection if the agent interpolates user-provided descriptions into these command strings without sufficient sanitization.
[REMOTE_CODE_EXECUTION]: The workflow involves the creation and immediate activation of new executable logic. This capability allows the agent to build and run code derived from user prompts, which presents a risk if the instructions are influenced by malicious user input.

create-plugin