The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to perform its core functions, including executing the complexa analysis CLI for scientific evaluation, a preflight.sh script for environment validation, and write_manifest.py for audit logging. These operations are well-scoped to the stated purpose of structure evaluation.
[INDIRECT_PROMPT_INJECTION]: The skill processes external biological data (PDB files) and parses metrics from generated CSV files to provide analysis summaries to the user. This data ingestion represents a standard attack surface for indirect prompt injection.
Ingestion points: External PDB files in a user-provided directory and generated *_results_*.csv files.
Boundary markers: No explicit delimiters or safety instructions are defined for parsing the scientific output.
Capability inventory: Execution of bioinformatics CLI tools and local manifest file generation.
Sanitization: No specific sanitization of metric strings is documented before summarizing results in the agent response.
[DYNAMIC_EXECUTION]: The skill references direct module invocation via python -m proteinfoundation.evaluate for debugging. This is a standard entry point for the tool's underlying Python modules and does not involve runtime code generation from untrusted sources.
[EXTERNAL_DOWNLOADS]: The skill includes references to source code and documentation hosted on NVIDIA's official GitHub repository. These references provide necessary context and configuration for the scientific tools.

complexa-evaluate-pdbs