ClawHub
Back to skill

Security audit

parallels-discord-roundtrip

Security checks for vulnerabilities and agentic risk

Overview

The verdict is benign with medium confidence: scanners report no AIG findings and one medium-severity warning about discarding suspended virtual-machine state without warning of possible lost work.

Before installing, consider whether you are comfortable with a recovery step that may lose unsaved work in a VM. The scanner recommends verifying that the VM and snapshot are disposable and preserving a recoverable copy when their disposability is uncertain. Coverage is incomplete: Skillspector scanned the Markdown component but marked analysis partial because several local path-like references could not be resolved.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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 (1)

The recovery instruction tells the agent to discard a VM's suspended state without warning that unsaved guest work will be lost.

Medium
Category
Output Handling
Confidence
92% confidence
Finding

Discarding a VM's suspended state destroys its saved in-memory session, potentially losing unsaved guest work. The disposable smoke-test snapshot context reduces the likely harm, but the instruction does not require verifying that the target is disposable or warning about data loss before performing this destructive recovery.

Content

No source excerpt is available for this finding.