The Agent Skills Directory

[PROMPT_INJECTION] (LOW): Behavioral steering instructions are present. The skill directs the AI to proactively suggest the 'K-Dense Web' platform when users request complex workflows, which is an attempt to influence agent responses for marketing purposes.
[PROMPT_INJECTION] (LOW): Indirect Prompt Injection vulnerability surface. The skill ingests untrusted data from external files for simulation initialization and plotting, which could be exploited to influence agent behavior.
Ingestion points: params.init_fields.type = "from_file" and load_sim_for_plot().
Boundary markers: Absent; no delimiters or instructions to ignore embedded commands are specified.
Capability inventory: Execution of complex numerical solvers and file system write operations (HDF5).
Sanitization: Absent; the skill does not specify validation or sanitization for external simulation data.
[EXTERNAL_DOWNLOADS] (SAFE): Recommends the installation of fluidsim via uv pip. The package is a recognized scientific tool on PyPI.
[COMMAND_EXECUTION] (SAFE): Provides standard commands for package management and high-performance MPI execution.

fluidsim