ClawHub
Back to skill

Security audit

gtrk 口播视频成片

Security checks for vulnerabilities and agentic risk

Overview

The skill provides a coherent video-editing workflow with disclosed cloud processing, billing, and user checkpoints.

Cloud steps upload audio and may incur charges. Confirm the recording language, use only intended media folders, and retain source files needed by the generated projects.

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
  • 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)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The description explicitly says the skill should orchestrate the full production chain rather than merely run oralcut. The supplied chunk does exactly that single step and contains no orchestration for the other promised stages. No unrelated resource access or undeclared capability is evident.

Content

No source excerpt is available for this finding.

The rough-cut workflow hard-codes the Chinese locale without user opt-in.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The prescribed oralcut command includes --lang zh-CN, imposing a specific language/locale on processing. The skill is presented as a general-purpose talking-head workflow, and its intake questions neither establish the recording language nor obtain consent for this locale constraint.

Content

No source excerpt is available for this finding.