The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from document files (HTML, Markdown, Text, RST) and review metadata (.review.json) to perform automated file edits, creating a surface for indirect prompt injection.
Ingestion points: Document source files and .review.json files ingested via the review-doc and apply-review commands.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are defined for the agent when processing document content or review annotations.
Capability inventory: The skill possesses file-write capabilities, as the apply-review command is designed to edit source files in place based on the review queue.
Sanitization: The instructions do not specify any sanitization, validation, or filtering of the review comments before they are interpreted and applied to the project files.

doc-review