The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of user-defined data through custom problem definitions, such as those demonstrated in scripts/custom_problem_example.py. These ingestion points allow users to provide objective functions and constraints for optimization. However, the capability inventory is limited to numerical computation and visualization using standard scientific libraries (NumPy, SciPy, Matplotlib). No dangerous capabilities such as network access, arbitrary command execution, or sensitive file operations were detected. Boundary markers are absent, but the nature of the data processing (numerical optimization) minimizes the risk of prompt-based exploitation. No sanitization logic was found, as the skill focus is on mathematical evaluation.
[EXTERNAL_DOWNLOADS]: The skill references the installation of the pymoo library and its dependencies (NumPy, SciPy, Matplotlib, Autograd) via standard package managers. These are well-known, trusted open-source libraries in the scientific computing ecosystem. No suspicious or unverified external sources were identified.
[COMMAND_EXECUTION]: The provided scripts utilize standard Python library imports and function calls for optimization tasks. No use of os.system, subprocess, or other dangerous shell execution methods was found. All operations are confined to the computational logic of the optimization algorithms.

pymoo