Improve package descriptions

scientific-packages/diffdock/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: diffdock
-description: This skill provides comprehensive guidance for using DiffDock, a state-of-the-art diffusion-based molecular docking tool that predicts protein-ligand binding poses. Use this skill when users request molecular docking simulations, protein-ligand binding predictions, virtual screening, structure-based drug design tasks, or need to predict how small molecules bind to protein targets. This skill applies to tasks involving PDB files, SMILES strings, protein sequences, ligand structure files, or batch docking of compound libraries.
+description: This skill provides comprehensive guidance for using DiffDock, a state-of-the-art diffusion-based deep learning tool for molecular docking that predicts 3D binding poses of small molecule ligands to protein targets. Use this skill when users request molecular docking simulations, protein-ligand binding pose predictions, virtual screening campaigns, structure-based drug design, lead optimization, binding site identification, or computational drug discovery tasks. This skill applies to tasks involving PDB protein structure files, SMILES ligand strings, protein amino acid sequences, ligand structure files (SDF, MOL2), batch docking of compound libraries, confidence score interpretation, ensemble docking with multiple protein conformations, integration with scoring functions (GNINA, MM/GBSA), parameter optimization for specific ligand types, troubleshooting docking issues, or analyzing docking results and ranking predictions. DiffDock predicts binding poses and confidence scores but NOT binding affinity - always combine with scoring functions for affinity assessment. Suitable for small molecule ligands (100-1000 Da), drug-like compounds, and small peptides (<20 residues), but NOT for protein-protein docking, large peptides, covalent docking, or membrane proteins without caution.
 ---
 
 # DiffDock: Molecular Docking with Diffusion Models