The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to execute shell commands directly on the host system via the Bash tool, specifically nb plugin list, nb plugin enable, and nb plugin disable.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data that could lead to unexpected behavior if not handled securely.
Ingestion points: The skill accepts plugins (a string array) and runtime_env_name as inputs in SKILL.md.
Boundary markers: There are no explicit delimiters or instructions to treat inputs as literal strings rather than executable fragments.
Capability inventory: The skill uses the Bash tool to run commands constructed from these inputs.
Sanitization: There is no mention of input validation, escaping, or sanitization for the plugin names or environment variables before they are interpolated into the shell commands.
[DYNAMIC_EXECUTION]: The skill dynamically assembles shell commands by concatenating base commands with user-supplied variables for plugin names and environment settings. While intended for administrative tasks, this pattern is a common vector for command injection if the underlying agent does not apply strict escaping to the arguments passed to the Bash tool.

nocobase-plugin-manage