The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes complex external data formats which can serve as a vector for indirect injection.
Ingestion points: The skill ingests PDB files for proteins, SMILES strings or SDF/MOL2 files for ligands, and batch processing instructions via CSV files as documented in SKILL.md and references/workflows.md.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore potential instructions embedded within the molecular data or CSV fields.
Capability inventory: The skill environment executes shell commands via Python and allows for file modifications via instructions.
Sanitization: The documentation does not describe any sanitization or validation performed on the structure files or CSV content before they are processed by the inference engine.
[EXTERNAL_DOWNLOADS]: The skill retrieves necessary model components from external sources.
Automatically downloads ESMFold model weights (esm2_t33_650M_UR50D) from Torch Hub into the ~/.cache/torch/hub/checkpoints directory during the first execution of sequence-only workflows.
[COMMAND_EXECUTION]: The skill involves the execution of shell commands to perform its tasks.
Provides instructions for running docking tasks using python3 -m inference.
Recommends using the sed utility to modify constants within the inference.py source code to adjust system resource limits for the sandbox.

diffdock