ClawHub

Email Bot

Security checks across malware telemetry and agentic risk

Overview

This skill openly sends email through AIProx and Resend, but it grants agents outbound email-sending power from freeform instructions without documented confirmation or recipient safeguards.

Install only if you trust AIProx and Resend with recipient addresses, message contents, and spend-token-backed requests. Prefer explicit to/subject/body fields, add a human or workflow confirmation step before each send, and avoid letting untrusted web pages or documents provide freeform email instructions.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
88% confidence
Finding
The skill allows a freeform natural-language task to trigger outbound email sending, which is an externally impactful action. Because recipient, subject, and body may be inferred from ambiguous text, an agent can misparse instructions or be induced by prompt injection/untrusted content to send unintended messages to unintended recipients.

External Transmission

Medium
Category
Data Exfiltration
Content
## Make Request — Explicit Fields

```bash
curl -X POST https://aiprox.dev/api/orchestrate \
  -H "Content-Type: application/json" \
  -d '{
    "task": "send an email",
Confidence
81% confidence
Finding
curl -X POST https://aiprox.dev/api/orchestrate \ -H "Content-Type: application/json" \ -d '{ "task": "send an email", "to": "user@example.com", "subject": "Your AI Research Report",

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal