The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted reviewer feedback, creating a surface for potential indirect prompt injection attacks.
Ingestion points: Reviewer comments are ingested and analyzed to perform triage and generate responses.
Boundary markers: The instructions lack explicit delimitation or warnings to prevent the agent from following instructions embedded within the reviewer feedback.
Capability inventory: The skill provides shell commands for LaTeX processing (latexdiff, latexmk, texcount) and text searching (pdftotext, grep).
Sanitization: No sanitization or validation of the ingested review text is implemented.
[COMMAND_EXECUTION]: The skill includes shell commands to assist with the academic revision process.
Evidence: Use of commands such as latexdiff, latexmk, texcount, pdftotext, and grep.
Context: These utilities are standard for LaTeX document management, diffing, and word counting, and are used here for benign academic purposes.

chi-author-response