The Agent Skills Directory

[SAFE]: The skill implements comprehensive input validation. The _common.py utility includes functions that explicitly reject network URLs and symbolic links for all input and output files, mitigating risks of remote file inclusion or unauthorized local file access.
[SAFE]: Robust resource management is enforced. The simulation scripts define a BoundedEnvironment that caps the number of processed events, entities, and simulation time, preventing potential denial-of-service through infinite loops or memory exhaustion.
[SAFE]: The skill follows security best practices for file operations. Reports and traces are written atomically using temporary files with restricted permissions (mode 0600), and JSON processing is hardened against non-standard constants and duplicate keys.
[SAFE]: Environment isolation is prioritized. The skill uses a pinned version of the SimPy library (4.1.2) and avoids all network activity or dynamic code execution during simulation runs, ensuring that model behavior remains local and predictable.
[SAFE]: Deterministic behavior is maintained through the use of stable, hash-derived seeds for random number generators, ensuring replicability without reliance on external entropy sources.

simpy