infrastructure-agent

Pass

Audited by Gen Agent Trust Hub on Sep 4, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill generates infrastructure configurations and executes validation scripts based on inputs from external project artifacts.
  • Ingestion points: Reads service maps, API contracts, and domain rules from contracts/ and the project README (SKILL.md).
  • Boundary markers: The instructions lack explicit delimiters or warnings to ignore instructions embedded within the ingested data files.
  • Capability inventory: Employs Bash for system probing and service validation, alongside Write and Edit for generating Dockerfiles, Makefiles, and CI/CD workflows (SKILL.md, references/validation-checklist.md).
  • Sanitization: No specific filtering or escaping mechanisms are mentioned for data values extracted from input files before they are interpolated into scripts or configurations.
  • [COMMAND_EXECUTION]: The skill utilizes the Bash tool for documented infrastructure tasks including environment verification using lsof to detect port collisions, managing local containers via docker compose, and testing service health with curl targeting localhost (references/validation-checklist.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 4, 2026, 02:32 PM