Back to skill

Security audit

agent-desktop-windows

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent Windows desktop automation guide with sensitive capabilities disclosed and scoped to its stated purpose.

Install this only if you want an agent to help with Windows desktop automation. Treat notification, clipboard, screenshot, and UI tree output as sensitive, use headed actions only when you intend foreground interaction, and leave the cursor overlay disabled on shared same-account machines or for users sensitive to motion.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger text includes a very broad condition such as 'or any Windows GUI interaction task,' which can cause this skill to activate for many unrelated requests. Over-broad invocation increases the chance that automation guidance is applied in the wrong context, potentially leading an agent to take GUI actions or prioritize this skill when a narrower, safer skill would be more appropriate.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

These lines explicitly say the overlay does not collapse when the OS accessibility preference disables animations, and that callers who reduced motion for accessibility reasons still see full travel animation. That is a natural-language policy concern because it forces a locale/accessibility behavior without offering user choice or opt-in at the behavior level.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
75% confidence
Finding

For markdown files, SQP-2 applies when behavior affecting privacy or system integrity is described without a clear warning. The section introduces a detached renderer process and control pipe at L097-L100, but the concrete risk that same-user processes can drive the overlay and display arbitrary text is only disclosed later at L120-L143, rather than as an upfront warning near the feature description.

Content

No source excerpt is available for this finding.