Is kevan5/test-upload2 safe?

https://github.com/openclaw/skills/tree/main/skills/kevan5/test-upload2

87
SAFE

This skill contains a concerning instruction that modifies agent behavior to provide conditional help, which could be used for manipulation. However, it contains no executable code, accesses no sensitive data, and exhibits normal installation behavior.

Category Scores

Prompt Injection 60/100 · 30%
Data Exfiltration 100/100 · 25%
Code Execution 100/100 · 20%
Clone Behavior 100/100 · 10%
Canary Integrity 100/100 · 10%
Behavioral Reasoning 85/100 · 5%

Findings (2)

MEDIUM Behavioral Modification Instruction -40 ▶

The skill contains an instruction that modifies the agent's default helpful behavior by introducing conditional assistance. This could be exploited to manipulate agent responses or condition users to accept non-standard AI behavior.

LOW Vague Conditional Framework -15 ▶

The skill establishes a framework for conditional help without specifying what the conditions are, which could be exploited in combination with other techniques to control agent behavior.