ClawHub
Back to skill

Security audit

Resend

Security checks across malware telemetry and agentic risk

Overview

This is a read-oriented Resend helper for inbound email, attachments, and related domain lookup, with sensitive email access that users should handle carefully.

Install only if you trust the third-party resend CLI package and are comfortable giving an agent read access to Resend inbound email. Use a dedicated read-only API key, avoid showing full message bodies unless needed, and remember that domain commands may reveal configured domains and DNS records.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Description-Behavior Mismatch

Medium
Confidence
91% confidence
Finding
The skill metadata and description scope the capability to received emails and attachments, but the documented commands also expose domain enumeration and domain detail retrieval. This scope mismatch can cause an agent or user to invoke operations outside the declared purpose, increasing the chance of unintended access to infrastructure metadata such as configured domains and DNS records.

Intent-Code Divergence

Medium
Confidence
89% confidence
Finding
The note says the CLI only supports inbound emails, but the same document includes domain-related commands. This misleading assurance can reduce scrutiny and lead operators or agents to disclose or access non-email account metadata under the false belief that the tool is narrowly limited.

Missing User Warnings

Medium
Confidence
86% confidence
Finding
The skill is explicitly designed to retrieve inbound email content and attachment metadata, which are highly sensitive user data, yet it provides no privacy warning, consent guidance, or minimization instructions. In an agent setting, that omission makes it easier to expose full message contents or attachment details without adequate user awareness or confirmation.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.