ClawHub
Back to skill

Security audit

Dlazy Superres

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed cloud image-upscaling skill with no evidence of hidden or malicious behavior, though users should understand that images and API credentials are handled by dLazy.

Install only if you are comfortable sending selected images or image URLs to dLazy's cloud service and storing or supplying a dLazy API key. Prefer `DLAZY_API_KEY` or verify local config permissions on shared machines, and use `--dry-run` when you want to inspect payload and cost before making a request.

SkillSpector

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

Intent-Code Divergence

Medium
Confidence
88% confidence
Finding
The skill is documented as an image super-resolution tool, but the example invokes `dlazy superres --prompt '提示词内容'`, even though the declared options only show `--image` and do not define `--prompt`. This inconsistency can mislead an agent into sending unintended free-form text to a hosted generation API or using the tool in a broader way than advertised, increasing the risk of unexpected remote processing and data disclosure.

Vague Triggers

Medium
Confidence
74% confidence
Finding
The trigger keyword `superres` is generic and lacks clear activation constraints, increasing the chance the skill will be invoked in unintended contexts. In an agent environment, broad triggers can cause accidental transmission of user-supplied image URLs or local file paths to third-party endpoints (`api.dlazy.com` and `files.dlazy.com`), creating privacy and data-handling risk even though the skill's purpose is legitimate.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.