infrastructure-operations
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill contains no executable code, network operations, or hardcoded credentials. It provides purely instructional guidance for infrastructure management, deployment, and operational diagnosis.
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read external files such as environment templates, deployment configurations, and documentation. This creates a surface for indirect prompt injection where instructions embedded in those project files could influence the agent's behavior.
- Ingestion points: Reads environment templates, deployment configurations, service definitions, documentation, and package/task scripts as specified in SKILL.md.
- Boundary markers: None specified in the instructions.
- Capability inventory: Uses repository wrappers, service CLIs, and container/orchestrator paths for state inspection and mutation (SKILL.md).
- Sanitization: No specific sanitization or filtering of external configuration data is mentioned.
Audit Metadata