microvm-nix

Pass

Audited by Gen Agent Trust Hub on Oct 10, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external NixOS configurations and flake inputs, which represents a potential attack surface for indirect instructions.
  • Ingestion points: NixOS configurations, flake inputs, and imported modules are read into the agent's context (SKILL.md, references/declaring.md).
  • Boundary markers: No specific delimiters or warnings to ignore instructions embedded in data are present.
  • Capability inventory: The skill provides guidance for executing administrative commands and building NixOS systems (references/interfaces.md, references/declarative.md).
  • Sanitization: No explicit sanitization of external configuration data is performed before processing.
  • [EXTERNAL_DOWNLOADS]: The skill fetches configuration modules from the project's official repository on GitHub.
  • Evidence: Uses builtins.fetchGit to download the host module from the microvm-nix GitHub organization (references/host.md).
  • [COMMAND_EXECUTION]: The documentation includes standard administrative commands for system configuration.
  • Evidence: References commands such as sudo ip tuntap add and sudo chown for managing network interfaces (references/interfaces.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 10, 2026, 03:53 PM