game-multiplayer

Pass

Audited by Gen Agent Trust Hub on Nov 9, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from .gameforge/project.json and the design/ folder to tailor its architectural advice. While this introduces an input surface for untrusted data, the skill is configured to output static documentation (ADRs and test plans) and explicitly instructs the agent to enforce server-side validation, significantly limiting the impact of potential injection attacks.
  • [COMMAND_EXECUTION]: Documentation within references/testing-and-cost.md includes shell commands for the tc utility (e.g., sudo tc qdisc) to simulate network conditions. These are presented as educational references for developers to perform manual network testing and are not intended for automated execution by the AI agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 9, 2026, 02:57 AM