The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized behaviors were detected across the skill files. The instructions are clearly focused on academic writing assistance and structural paper diagnosis.- [EXTERNAL_DOWNLOADS]: The skill is configured to retrieve data from well-known academic sources, such as OpenReview and official conference proceedings. These operations are used to extract high-level patterns from accepted research papers and are consistent with the skill's stated purpose.- [PROMPT_INJECTION]: The skill processes untrusted external data from websites and research papers, which constitutes an indirect prompt injection surface. The risk is inherent to the use-case, and no exploitation or malicious instructions were found in the skill code.
Ingestion points: WebFetch and WebSearch tools used to access conference websites and external PDFs.
Boundary markers: The skill does not explicitly define delimiters for untrusted data.
Capability inventory: Access to Bash, Write, and Edit tools for local file manipulation.
Sanitization: No explicit sanitization of web-fetched content is mentioned.

conference-writing-adapter