The Agent Skills Directory

[COMMAND_EXECUTION]: The skill generates and executes shell commands using the official claude plugin utility (install, uninstall, update) and performs directory deletion for cache maintenance. These actions are triggered based on the state of the local filesystem. The instructions explicitly mandate user confirmation before proceeding with mutative or destructive workflows.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface (Category 8) by processing untrusted metadata from source repositories and local directories to determine management actions.
Ingestion points: Data is read from ~/.claude/plugins/cache/, ~/.claude/skills/, and .claude-plugin/marketplace.json across all defined workflows.
Boundary markers: No explicit delimiters or instructions to ignore embedded content are used when these inputs are interpolated into the agent's context.
Capability inventory: The skill has the capability to run plugin management commands and delete filesystem directories.
Sanitization: There is no evidence of sanitization or strict validation of the metadata (e.g., plugin names or versions) before they are used as command arguments.

skill-governance