The Agent Skills Directory

[METADATA_POISONING]: The skill metadata identifies 'K-Dense Inc.' as the author, which is inconsistent with the user 'leonchaox'. Furthermore, the skill cites a research paper with an arXiv ID (2510.16234) corresponding to October 2025; this future-dated reference suggests the citation may be fabricated or deceptive to lend false authority to the skill.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to evaluate untrusted scholarly work (papers, proposals, literature reviews), which serves as a potential vector for indirect prompt injection. 1. Ingestion points: The evaluation workflow in SKILL.md (Step 1) defines the ingestion of various types of scholarly work into the agent's context. 2. Boundary markers: There are no explicit delimiters or instructions provided to the agent to isolate or ignore instructions that might be embedded within the ingested text. 3. Capability inventory: The skill possesses the capability to execute local Python scripts and write reports/images to the filesystem. 4. Sanitization: No validation, filtering, or sanitization is performed on the ingested content before it is processed.
[COMMAND_EXECUTION]: The skill relies on local script execution for its core functionality. While scripts/calculate_scores.py is provided and appears benign, scripts/generate_schematic.py is referenced for automated diagram generation but is not included in the skill's file list, representing an unverifiable command execution surface.

scholar-evaluation