The Agent Skills Directory

[SAFE]: The skill provides domain-specific instructions for academic publishing, including journal positioning, topic preferences, and structural advice. It does not contain any executable code, obfuscation, or malicious instructions.
[DATA_EXFILTRATION]: The skill includes instructions to read internal relative paths such as ../../resources/source-basis.md and ../../resources/official-source-map.md. These are localized resource files used to anchor the agent's knowledge to official journal sources and do not represent exposure of sensitive system files or user credentials.
[PROMPT_INJECTION]: The skill directs the agent to search for and ingest external data from official journal websites to verify submission guidelines. This constitutes an indirect prompt injection surface; however, the scope is limited to academic formatting and procedural verification, which is a low-risk application with no access to sensitive capabilities or dangerous commands.

management-world