sol-terra-luna-auto-router
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill instructions define clear boundaries and prohibit the agent from overriding user-defined constraints, such as model locks. It uses recursion markers to prevent infinite routing loops.
- [DATA_EXPOSURE]: The instructions mandate data minimization, requiring the agent to use compact briefs and specific file paths instead of passing full conversation contexts to sub-tasks.
- [REMOTE_CODE_EXECUTION]: Installation scripts are localized and use standard system utilities. The routing mechanism relies on native platform tools for task management.
- [COMMAND_EXECUTION]: Orchestration features utilize intended CLI tools within the execution environment without introducing unauthorized or dangerous bypasses.
- [CREDENTIALS_SAFE]: No hardcoded secrets, environment variables, or private keys were detected in any of the skill files.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user requests (SKILL.md) and processes them through various adapters. It implements boundary markers via structured handoff briefs (Outcome, Inputs, Constraints) and recursion markers for child tasks. The capability inventory includes thread forking and agent spawning. Sanitization is achieved through strict model-routing policies and explicit importance-aware delegation templates.
Audit Metadata