The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and systematically process untrusted external data in the form of thesis manuscripts (Markdown, Word exports, or .tex files).
Ingestion points: Documents provided by the user or read via file system operations during the three-round review process described in SKILL.md.
Boundary markers: The instructions do not define specific delimiters or "ignore embedded instructions" protocols to prevent the agent from following malicious commands hidden within a thesis manuscript.
Capability inventory: The skill uses file reading, keyword scanning, and complex summarization capabilities to generate reports.
Sanitization: No input validation or sanitization of processed text is specified.
[COMMAND_EXECUTION]: The skill references the potential execution of local scripts such as scripts/ai_detector.py and scripts/text_analyzer.py for automated text analysis. These are documented as internal vendor tools and do not involve remote code downloads from unknown sources.

thesis-reviewer