fluidsim

Pass

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill demonstrates dynamic generation of Python scripts from templates to automate parametric studies on HPC clusters as shown in references/advanced_features.md. It also provides instructions for parallel execution using mpirun and interacting with cluster submission systems.
  • [PROMPT_INJECTION]: The skill describes workflows for ingesting external simulation data which presents a surface for indirect prompt injection. 1. Ingestion points: Simulation directories, HDF5 (.h5), and text files loaded via load_sim_for_plot. 2. Boundary markers: No markers or 'ignore' instructions are present in the provided examples. 3. Capability inventory: Capability to write generated scripts to the file system and submit them for execution on cluster nodes. 4. Sanitization: No sanitization or validation of external data is demonstrated before it is used to parameterize generated scripts.
  • [EXTERNAL_DOWNLOADS]: The skill recommends installing the fluidsim package and its dependencies (fluidfft, pyfftw, mpi4py) using the uv package manager. A repeated command typo uv uv pip is present in the main instructions but corrected in reference files.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 13, 2026, 10:21 AM