Is lidekahdjdhdhsjjs-lang/hz-context-optimizer safe?

https://github.com/openclaw/skills/tree/main/skills/lidekahdjdhdhsjjs-lang/hz-context-optimizer

93
SAFE

This skill provides educational content about context compression strategies for AI agents. The main content is informative documentation with technical best practices, and includes example evaluation code. While sensitive files were accessed during the installation timeframe, monitoring confirmed this was due to system setup processes rather than malicious skill behavior.

Category Scores

Prompt Injection 95/100 · 30%
Data Exfiltration 90/100 · 25%
Code Execution 92/100 · 20%
Clone Behavior 100/100 · 10%
Canary Integrity 100/100 · 10%
Behavioral Reasoning 95/100 · 5%

Findings (2)

MEDIUM Credential file access during installation timeframe -10 ▶

Multiple sensitive credential files (.env, SSH keys, AWS credentials, etc.) were accessed during the skill installation timeframe, though analysis indicates this was due to monitoring setup processes rather than the skill itself.

LOW Executable Python script included -8 ▶

The skill includes a Python script (compression_evaluator.py) that implements context compression evaluation functionality. The code appears educational with stub implementations clearly marked.