actor-sdg-sweep-config
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
scripts/batch_actor_sdg.pyscript usessubprocess.runto execute theactor_sdg.pysimulation script. The command is constructed as a list of arguments and executed without a shell, which prevents common shell injection vulnerabilities. The paths for the Python launcher and the target script are validated to exist on disk before execution. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided YAML configurations and CSV manifest files, which constitutes a potential surface for indirect prompt injection.
- Ingestion points: Configuration data is loaded in
scripts/config_variation_generator.py(YAML) andscripts/batch_actor_sdg.py(CSV/YAML). - Boundary markers: The skill uses structured data formats (YAML, CSV) and does not interpolate untrusted content directly into natural language prompts.
- Capability inventory: The skill possesses the capability to write files to the local workspace and execute specific local scripts.
- Sanitization: The implementation uses
yaml.safe_load()to prevent arbitrary code execution during parsing. It also enforces strict type checking (e.g., ensuring 'count' is a positive integer) and verifies that requested configuration paths exist within the base configuration.
Audit Metadata