The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it interpolates user-provided engineering parameters directly into executable templates without explicit sanitization.
Ingestion points: Engineering parameters such as dimensions, flow rates, and heat loads provided by the user, as defined in the parameters sections of template files (e.g., references/openfoam-templates/data-center-airflow.yaml).
Boundary markers: No explicit delimiters or boundary markers are instructed to the agent to distinguish between user data and template instructions.
Capability inventory: The skill produces SimulationPackage objects containing shell command sequences (run_instructions), Python scripts (FreeCAD macros), and interactive UI code (React artifacts).
Sanitization: There is no evidence of input validation or escaping logic to prevent malicious parameters from altering the structure of the generated executable files.
[COMMAND_EXECUTION]: The skill generates shell command sequences intended for local execution by the user, including solvers like simpleFoam and mesh tools like blockMesh, which are delivered via the run_instructions field.
[DYNAMIC_EXECUTION]: The skill generates script-like content including FreeCAD Python macros and React TSX components from predefined templates based on user inputs, representing a dynamic code generation workflow.

simulation-bridge