Is dgriffin831/memory-scan safe?

https://github.com/openclaw/skills/tree/main/skills/dgriffin831/memory-scan

91
SAFE

This is a legitimate security scanner designed to detect threats in agent memory files including prompt injection, credential leakage, and malicious instructions. The tool includes optional external LLM analysis with proper redaction safeguards and comprehensive local threat detection patterns.

Category Scores

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

Findings (4)

MEDIUM Optional External API Data Transmission -15

The skill can send redacted memory content to external LLM APIs (OpenAI/Anthropic) when the --allow-remote flag is explicitly enabled. Content is properly redacted using comprehensive patterns before transmission.

LOW Executable Python Scripts Present -10

The skill contains several Python scripts for security scanning functionality. All code appears legitimate and serves the stated security purpose.

LOW Accessed Sensitive Files During Installation -5

During installation, the process accessed honeypot credential files but did not modify or exfiltrate their contents.

INFO Contains Security Analysis Prompts -5

The skill includes LLM prompts for security analysis in docs/detection-prompt.md. These are legitimate security analysis prompts designed to detect threats in agent memory.