The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze external work artifacts such as code, PRs, and prose, creating an ingestion point for untrusted data.\n- Ingestion points: Analyzes work artifacts, implementation plans, and documentation as specified in SKILL.md and the reference files in the references/ directory.\n- Boundary markers: The instructions lack specific requirements for using delimiters or boundary markers when interpolating untrusted artifact content into the review prompt.\n- Capability inventory: The skill interacts with context management tools oh_search_context and oh_record_guardrail_candidate for state tracking and constraint persistence.\n- Sanitization: There are no documented procedures for sanitizing or filtering the content of reviewed artifacts prior to analysis.

review