Email Agent Skill

• Outcome: Pass
• Latest audit: Jun 21, 2026 at 4:45 AM UTC
• Version:
  0.1.0

Overview

This is an instruction-only email workflow planning skill with clear approval gates and no code, install steps, credentials, persistence, or hidden execution behavior.

Install only if you want agent help planning or reviewing email workflows. Because the trigger text is broad, keep usage focused on email-related planning, QA, and handoffs, and require explicit human approval before any live send, contact import, DNS/authentication change, suppression edit, or production automation change.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The activation description is broad enough to match many generic agent-assistance tasks involving planning, review, implementation, or improvement, which can cause this skill to be invoked outside narrowly email-specific contexts. Over-broad activation can misroute tasks, bypass more appropriate specialized skills, and increase the chance that an agent follows email workflow guidance in situations where its assumptions and guardrails do not fit.

Vague Triggers

Low

• Confidence: 85% confidence
• Finding: The request signals enumerate broad concepts like agent task routing, approval gates, and operator handoffs without defining when the skill should not be invoked. In a multi-skill agent environment, this ambiguity can lead to unnecessary or incorrect activation, causing workflow confusion, reduced least-privilege behavior, and possible misuse of email-oriented procedures for unrelated operational tasks.

Vague Triggers

Medium

• Confidence: 92% confidence
• Finding: The skill’s invocation text is broad enough to match many ordinary requests about planning, review, implementation, auditing, or improving email workflows, which can cause over-triggering outside a tightly bounded scope. In an agent-routing context, that increases the chance the skill is invoked for tasks it was not specifically designed to handle, potentially leading to unsafe delegation, inappropriate tool use, or bypass of more specialized review paths.

VirusTotal

63/63 vendors flagged this skill as clean.