compliance-readiness
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Python scripts such as framework_selector.py and audit_simulator.py found within the project's scripts directory. This is consistent with its intended use as an automated compliance assessment tool.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external data from JSON files (profile.json, program.json, scope.json) to inform its analysis. This creates an attack surface for indirect prompt injection as these files are used as context for the agent's operations. 1. Ingestion points: profile.json, program.json, and scope.json in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: Execution of local Python scripts via subprocess. 4. Sanitization: Not explicitly handled in the skill instructions.
Audit Metadata