ClawHub
Back to skill

Security audit

arena-council

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly routes prompts to local LLMs, but it also automatically uses a “god mode” prompt-wrapper to bypass refusals and can load unreviewed helper code from another skill directory.

Install only if you intentionally want local multi-model routing and are comfortable disabling or reviewing the god-mode integration. Before use, inspect any sibling god-mode skill/code, consider removing that directory or editing the skill to prevent automatic prompt wrapping/probing, and avoid sending sensitive prompts unless you trust the local LM Studio setup and all selected models.

Static analysis

No suspicious patterns detected.