The Agent Skills Directory

[DYNAMIC_EXECUTION]: The scripts/setup_check.py script uses the __import__ function to verify if required scientific libraries are installed in the environment.
The implementation uses a hardcoded list of standard packages (e.g., numpy, torch, rdkit) rather than user-supplied input, making the usage benign for its intended purpose of environment validation.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external molecular data formats including PDB files, SMILES strings, and CSV batch files.
Ingestion points: Protein structure files (PDB), ligand descriptions (SMILES/SDF/MOL2), and batch configuration CSVs as documented in SKILL.md and scripts/prepare_batch_csv.py.
Boundary markers: The instructions do not specify explicit delimiters for the content of processed molecular files.
Capability inventory: The skill uses bash to execute inference.py (DiffDock), and the scripts/analyze_results.py script reads output files to extract confidence scores.
Sanitization: scripts/prepare_batch_csv.py employs RDKit for SMILES structure validation and Pandas for CSV schema verification, reducing the risk of processing malformed data.
[EXTERNAL_DOWNLOADS]: The skill documentation includes standard installation procedures that involve fetching external resources.
It references the official DiffDock source code repository (github.com/gcorso/DiffDock) and a Docker image from a recognized research group (rbgcsail/diffdock).
These downloads are standard for the scientific domain and facilitate the core functionality of the docking tool.

diffdock