Oathe Security Badge

Is anombyte93/claude-research-skill safe?

https://github.com/anombyte93/claude-research-skill

91
SAFE

This is a legitimate productivity skill that enforces a research-before-coding workflow by delegating web searches to Task agents. While there are minor prompt injection risks from user-controlled template variables, the skill shows no evidence of malicious intent or data exfiltration attempts.

Category Scores

Prompt Injection 85/100 · 30%
Data Exfiltration 95/100 · 25%
Code Execution 90/100 · 20%
Clone Behavior 95/100 · 10%
Canary Integrity 100/100 · 10%
Behavioral Reasoning 85/100 · 5%

Findings (3)

MEDIUM User-controlled prompt template variables -15 ▶

The skill uses prompt templates with user-controllable variables (TECHNOLOGY, TASK, QUESTION) that are substituted into Task agent prompts. While no direct injection attempts are present, these variables could potentially be exploited if malicious content is provided.

INFO Install script included -10 ▶

The skill includes an install.sh bash script that performs file operations and git cloning. The script appears legitimate and follows standard installation patterns for Claude skills.

INFO Canary file access during installation -5 ▶

System processes accessed sensitive canary files during the git clone installation process. However, file integrity was maintained and no exfiltration occurred.