The Agent Skills Directory

[SAFE]: The skill consists exclusively of markdown instructions and YAML configuration defining a specialized AI persona. No executable scripts, network operations, or file system interactions are included in the provided files.
[SAFE]: References to external scientific databases (e.g., ChEMBL, PubChem, SciFinder) and modeling software (e.g., Schrödinger, RDKit, OpenEye) are provided as contextual domain information for the agent's persona. These references do not involve automated downloads, installations, or remote code execution.
[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to process data from external sources such as research papers, patent databases, and lab notebooks (ELNs). While processing untrusted external data represents a potential surface for indirect prompt injection, the skill does not contain the code to perform these fetches, and the instructions are focused on professional scientific analysis. The risk is inherent to the described role and considered low.

medicinal-chemist