ClawHub
Back to skill

Security audit

Email Triage

Security checks for vulnerabilities and agentic risk

Overview

This instruction-only email triage skill is coherent and limited, though users should know it may inspect email metadata and short previews when invoked.

Install this only if you are comfortable with an agent using existing email access to review recent unread message metadata and short previews for priority senders. Use explicit prompts like "triage my inbox" to avoid accidental activation from vague requests.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (2)

Vague Triggers

Medium
Confidence
96% confidence
Finding
The invocation phrases include broad natural-language triggers such as 'what's important' and 'check for urgent email,' which can cause the skill to activate in situations where the user did not clearly intend inbox triage. In an email-handling context, accidental activation can expose message metadata or lead the agent to access sensitive inbox contents based on ambiguous requests.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The skill instructs the agent to fetch unread email metadata and then body previews for selected messages without requiring an explicit user warning or confirmation that inbox contents will be accessed. Because email commonly contains highly sensitive personal, financial, legal, or business data, this can result in privacy-invasive access or summarization beyond what the user reasonably expected from a vague triage request.

Static analysis

No suspicious patterns detected.