Back to skill

Security audit

agent-desktop

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marked this desktop skill suspicious because its guidance may change the active desktop and enables a Chromium debugging endpoint without enough access safeguards; confidence is medium.

Review the desktop-focus behavior and the Chromium debugging setup before installing. Consider whether you are comfortable with a workflow that may interrupt your desktop and could expose authenticated app access; the scan's partial coverage means some referenced material was not resolved.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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
Findings (2)

A command classified as observation-only opens an OS surface and changes foreground focus.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

Line 3 promises that these commands read UI state and never change it, but the executable example at line 119 invokes open-system-surface. Line 122 explicitly states that this command raises an OS shell surface and takes the foreground, contradicting the observation-only guarantee and potentially disrupting the user's active desktop context.

Content

No source excerpt is available for this finding.

The Chromium workflow enables a debugging endpoint without warning about access to authenticated app data.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The workflow recommends exposing a CDP endpoint on an authenticated desktop application without explaining its security implications or requiring access restrictions. A process able to connect could potentially read authenticated application content or perform actions as the user; the supplied documentation does not establish whether the endpoint is reachable beyond the local machine.

Content

No source excerpt is available for this finding.