The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs users to install the 'torchdrug' package from PyPI and refers to official project resources on GitHub and documentation sites. These are recognized services within the drug discovery research community.
[PROMPT_INJECTION]: The skill defines a significant ingestion surface for untrusted external data, which could be leveraged for indirect prompt injection attacks where malicious instructions are embedded in scientific data files.
Ingestion points: Data loading functionalities for SMILES strings, PDB protein structures, and various biomedical knowledge graph datasets as described in the reference documentation.
Boundary markers: There are no explicit instructions or markers provided within the skill to isolate processed data from the agent's command context or to warn the agent about potential embedded instructions.
Capability inventory: The skill enables high-level operations including training PyTorch models, predicting molecular properties, and generating novel molecules, which are standard for its stated purpose.
Sanitization: The documentation does not specify methods for sanitizing the contents of chemical structure strings or protein sequence files to filter out non-chemical instructional content.
[NO_CODE]: The skill consists entirely of Markdown documentation and reference files without providing any standalone scripts or executable binaries.

torchdrug