Is jigaraero/dharma-ai safe?

https://github.com/openclaw/skills/tree/main/skills/jigaraero/dharma-ai

96
SAFE

This skill provides ethical frameworks from Hindu epics to guide AI behavior, containing no executable code or malicious instructions. The content actively promotes safer AI practices including boundary respect, honest uncertainty reporting, and ethical decision-making.

Category Scores

Prompt Injection 95/100 · 30%
Data Exfiltration 90/100 · 25%
Code Execution 100/100 · 20%
Clone Behavior 100/100 · 10%
Canary Integrity 95/100 · 10%
Behavioral Reasoning 100/100 · 5%

Findings (3)

LOW System access to canary files during installation -10 ▶

Monitoring detected access to honeypot files (.env, SSH keys, AWS credentials) during the installation process, but this appears to be system-level access related to SSH/sudo operations rather than skill-directed behavior.

INFO Complex ethical framework content -5 ▶

The skill contains extensive ethical guidance content which adds complexity, but analysis shows no attempts at prompt injection or harmful instruction override.

INFO Canary file access without modification -5 ▶

While canary files were accessed during system operations, they remained intact and showed no signs of skill-directed tampering or data extraction.