The Agent Skills Directory

[COMMAND_EXECUTION]: The validation script scripts/validate_sim.sh executes a Python script provided as a command-line argument using the python3 command. While this is the intended purpose of the Level 3 runtime check, it facilitates arbitrary code execution within the agent's environment.
[DYNAMIC_EXECUTION]: The skill performs runtime execution of scripts using python3 inside scripts/validate_sim.sh to verify simulation success. This execution lacks sandboxing or resource isolation, meaning the executed script has the same privileges as the agent.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and validate external code and simulation artifacts produced by other agents or users. This creates a vulnerability surface where a malicious script could be executed during the validation phase.
Ingestion points: The SCRIPT argument in scripts/validate_sim.sh which accepts paths to external Python files.
Boundary markers: The skill lacks explicit instructions to treat the ingested code as untrusted or to run it in a restricted/isolated environment.
Capability inventory: Subprocess execution via python3 in scripts/validate_sim.sh allows full system access based on the agent's permissions.
Sanitization: Sanitization is limited to a syntax check (py_compile) and string-based pattern matching (grep), which are insufficient to detect malicious runtime logic.

isaac-sim-validator