The Agent Skills Directory

[SAFE]: The skill implements a legitimate administrative workflow for code refactoring. It includes validation steps to ensure it is operating on a valid plugin structure before initiating deeper analysis.
[INDIRECT_PROMPT_INJECTION]: The skill operates on a user-provided file path and processes the contents of that directory. This creates a potential surface for indirect prompt injection where instructions hidden within the plugin being refactored could attempt to influence the behavior of the orchestrating agent or downstream specialized agents.
Ingestion points: The workflow reads from .claude-plugin/plugin.json and the skills/ directory within the provided path.
Boundary markers: The orchestration instructions do not specify explicit delimiters or 'ignore' instructions for the content being analyzed.
Capability inventory: The skill has the capability to trigger further analysis and implementation tasks through plugin-creator:assessor and plugin-creator:implement-refactor, which involve file system operations.
Sanitization: No explicit sanitization of the plugin file content is performed at this orchestration level.

refactor-plugin