z-lua-core
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a protocol for ingesting and executing untrusted Lua scripts. While it mandates defensive measures, the ingestion of external script data represents a potential attack surface for indirect prompt injection if the host environment is not properly isolated.
- Ingestion points: Mentions execution of "untrusted or semi-trusted script content" in
SKILL.md. - Boundary markers: Recommends allowlisting the environment and stripping the standard library (e.g.,
os,io) to prevent host-level capability exposure. - Capability inventory: Discusses host API surface exposure and resource budgeting (CPU/Memory) via hook-based instruction counting.
- Sanitization: Advises against string-concatenation for script generation and recommends typed host-API boundaries to prevent code injection.
- [DYNAMIC_EXECUTION]: The skill describes patterns for runtime execution of Lua code using embedding libraries such as
mluaandgopher-lua. It explicitly highlights the risks associated with dynamic code generation from external inputs and provides guidance on avoiding unsafeload()patterns.
Audit Metadata