The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill interprets content from other skill files, creating an attack surface for embedded malicious instructions. * Ingestion points: Reads skill bodies and reference files using skill_view as seen in SKILL.md. * Boundary markers: No delimiters or ignore-instructions are used when processing external content. * Capability inventory: Uses skill_manage for file creation and modification, plus skill_view and skills_list for inspection. * Sanitization: External content is not sanitized before processing.
[COMMAND_EXECUTION]: The skill specifies the use of the omh utility for runtime status tracking. * Evidence: The skill instructs execution of omh runtime record commands within its workflow documentation.

omh-skill