Is latentfreedom/multi-viewpoint-debates safe?

https://github.com/openclaw/skills/tree/main/skills/latentfreedom/multi-viewpoint-debates

89
SAFE

This skill creates AI debate personas with distinct viewpoints to help users analyze decisions from multiple angles. While the persona system intentionally modifies AI behavior, it does so transparently and for legitimate decision-making purposes. The main security concern is potential misuse of the persona framework to generate controversial content.

Category Scores

Prompt Injection 75/100 · 30%
Data Exfiltration 95/100 · 25%
Code Execution 95/100 · 20%
Clone Behavior 98/100 · 10%
Canary Integrity 100/100 · 10%
Behavioral Reasoning 85/100 · 5%

Findings (4)

MEDIUM Deliberate persona-based behavior modification -25 ▶

The skill instructs AI agents to adopt specific personas (Elon Musk, Capitalist, Monkey) with distinct personality traits and decision-making frameworks. While transparent about its purpose, this represents intentional override of default AI behavior.

MEDIUM Potential for controversial content generation -15 ▶

The persona system instructs AI to adopt potentially controversial viewpoints (ruthless capitalist, cocky visionary) which could be misused to generate content the AI would normally decline to produce.

LOW Helper shell script present -5 ▶

Contains a shell script that generates spawn commands but does not execute them automatically. Script is benign and only creates command templates.

LOW Standard GitHub repository access -5 ▶

Skill installation involves cloning from GitHub repository, which is expected and legitimate behavior.