The Agent Skills Directory

[SAFE]: The skill implements a declarative manifest and on-demand reference loading, ensuring a controlled environment for AI reasoning based strictly on user-provided manuscript data.
[SAFE]: No network activity, subprocess execution, or suspicious file operations were found in the instructions or scripts.
[SAFE]: Analysis for obfuscation, including Base64, homoglyphs, and hidden characters, yielded no results.
[SAFE]: The skill includes explicit boundary checks and QA protocols that instruct the agent to avoid inventing information and to stay within the role of a reviewer.

nature-reviewer