The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is primarily designed to ingest and parse full-text research papers (XML, LaTeX, PDF text) from external academic repositories. As the skill's purpose is to process external scholarly data, there is a theoretical surface for indirect prompt injection if an article contained malicious instructions. However, the skill provides structured parsing logic (e.g., JATS XML parsing in open-access-mining-guide/SKILL.md) which inherently mitigates risk by segmenting content, and this ingestion is central to the tool's primary research function.
[EXTERNAL_DOWNLOADS]: The skill facilitates data retrieval from established scientific and research infrastructure, including the National Center for Biotechnology Information (NCBI), Cornell University's arXiv, the Harvard Dataverse, and the Open Science Framework (OSF). These downloads target official, well-known academic domains and repositories.
[CREDENTIALS_UNSAFE]: For services requiring authentication (such as the CORE API), the instructions specifically advise users to store API keys in environment variables rather than hardcoding them, demonstrating adherence to secure secret management practices.
[COMMAND_EXECUTION]: The documentation includes transparent examples of curl and wget commands to demonstrate how to interact with research APIs. These are provided for instructional purposes and do not involve obfuscated or unauthorized execution.

fulltext-skills