The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and refactor external UI code provided by users. This creates a surface for indirect prompt injection if the code contains instructions disguised as comments or data.
Ingestion points: Target UI files and components provided by the user for refactoring.
Boundary markers: The skill relies on a 'Safety gate' requiring characterization tests and a 'Preview' phase to plan changes before they are applied, acting as a functional boundary.
Capability inventory: The skill possesses the capability to write/edit files and execute shell commands for testing and delegation.
Sanitization: No specific input sanitization is mentioned; however, the impact is mitigated by the mandatory behavior-locking tests and manual review steps.
[COMMAND_EXECUTION]: The skill provides detailed instructions for invoking external command-line interfaces (CLIs) such as Claude Code (claude), Codex (codex), and a specialized utility omh. It includes specific guidance on managing permissions (--permission-mode), session persistence, and runtime recording. These commands are integral to the skill's function as a handoff guide for refactoring tasks.

omh-frontend-refactor