Back to skill
Skillv1.4.0
VirusTotal security
Agent Mail Guard — Email Sanitizer for AI Agents · External malware reputation and Code Insight signals for this exact artifact hash.
Scanner verdict
BenignApr 30, 2026, 4:42 AM
- Hash
- eff5397c1b973f8639b3891b2544ead04470f361d84d2b93e70be09a1334b5fa
- Source
- palm
- Verdict
- benign
- Code Insight
- Type: OpenClaw Skill Name: agent-mail-guard Version: 1.4.0 This skill bundle is designed to enhance the security of AI agents by sanitizing email and calendar content to prevent prompt injection and data exfiltration. All files, including the `SKILL.md` instructions, `sanitize_core.py` (containing extensive detection patterns and stripping functions), `sanitizer.py`, `cal_sanitizer.py`, and shell scripts, consistently demonstrate a clear defensive purpose. There is no evidence of intentional harmful behavior such as credential theft, unauthorized data exfiltration to external attacker-controlled endpoints, persistence mechanisms, or remote control. The `gog` CLI is used for fetching data, and all processing is local. The documentation explicitly warns agents against executing commands or visiting URLs based on email content, reinforcing its security-enhancing role. The only minor flaw is the silent failure of audit logging in shell scripts (`2>/dev/null || true`), which is a robustness issue, not a malicious act.
- External report
- View on VirusTotal