ClawHub
Back to skill

Security audit

Openclip

Security checks for vulnerabilities and agentic risk

Overview

OpenClip is a disclosed remote media-processing skill, with no executable payloads or hidden behavior found.

Install this only if you are comfortable connecting your agent to OpenClip and sending selected videos, images, URLs, or briefs to OpenClip for processing. Be especially careful with private media because completed outputs may be stored as service/CDN URLs.

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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)

Vague Triggers

Medium
Confidence
93% confidence
Finding
The skill declares very broad trigger phrases such as "transcribe this", "compress this video", "remove the background", and especially the generic "openclip", which overlap with common user requests across many contexts. This can cause the agent to invoke the OpenClip skill unexpectedly, sending user-provided media URLs or uploaded files to an external service without sufficiently explicit user intent, increasing the risk of unintended data disclosure or action hijacking.

Static analysis

No suspicious patterns detected.