The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from external sources, specifically peer reviewer comments and user-provided manuscript drafts.
Ingestion points: The skill ingests external text for revision and response generation in SKILL.md and references/revision-response.md.
Boundary markers: No specific delimiters or safety instructions are provided to help the agent distinguish between user instructions and external content during the revision process.
Capability inventory: The skill is strictly for text generation and does not have access to tools for file system modification, network communication, or code execution, which eliminates the impact of potential injections.
Sanitization: No input validation or sanitization is performed on the ingested text.
[SAFE]: The skill consists of markdown instructions and templates without any executable logic or remote dependencies. The mentioned sister skill, manuscript:humanizer, is a text-processing utility from the same author intended for academic formatting and does not introduce security risks.

manuscript-writing