Back to skill

Security audit

Persona Exec Assistant

Security checks for vulnerabilities and agentic risk

Overview

This executive-assistant skill is coherent, but it can send email from the user's workspace without clearly requiring approval first.

Review this skill before installing in a real executive workspace. It is suitable only if the agent is allowed to read Google Workspace data, and it should be amended or used with a strict approval rule before any email is sent.

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
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
Category
Not specified by scanner
Confidence
92% confidence
Finding

The manifest description says "Manage an executive's schedule, inbox, and communications," which is a very broad natural-language capability statement that overlaps with common assistant requests. The file does not provide explicit trigger phrases, exclusions, or scope limits that would help prevent unintended invocation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill instructs the agent to draft replies with gws gmail +send, which implies external communication, but it does not require explicit user confirmation or a warning before sending. In an executive-assistant context, this is more dangerous because messages may be sent to leadership, direct reports, or external contacts, increasing the risk of unauthorized or mistaken communications with reputational or operational impact.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.