ClawHub
Back to skill

Security audit

Dlazy Jimeng T2i

Security checks across malware telemetry and agentic risk

Overview

The skill is a disclosed dLazy text-to-image wrapper, but it merits Review because it stores an API key locally and can send broad piped input or local files to dLazy with limited guardrails.

Review before installing if the dLazy API key or generated prompts are sensitive. Prefer explicit invocation, consider npx or dry-run first, avoid @stdin/@* unless you know exactly what is being forwarded, pass only intended files, and check permissions on ~/.dlazy/config.json after login.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Vague Triggers

Medium
Confidence
88% confidence
Finding
The trigger keywords include broad everyday phrases such as '生成图片、文生图' and especially '画一张图', which can cause the skill to activate for ordinary user requests that did not explicitly intend to use this external SaaS tool. In this skill, mis-triggering is more dangerous because activation can lead to authenticated API calls, possible local file upload behavior, and billable usage against the user's dLazy account.

Vague Triggers

Medium
Confidence
95% confidence
Finding
The trigger phrases include broad, everyday language such as 'generate image, text to image' and 'draw a picture', which can cause the skill to be invoked unintentionally in ordinary conversations. Because this skill installs or runs a third-party CLI and may transmit prompts, local file paths, and authentication-backed requests to external services, accidental invocation can lead to unintended network access, data disclosure, or unexpected charges.

Ssd 3

Medium
Confidence
95% confidence
Finding
The skill explicitly documents that flags can consume '-', '@stdin', '@stdin:path', '@*', and upstream outputs, enabling an agent to pass entire envelopes or prior tool outputs directly into command arguments. In this context, those arguments are sent to dLazy APIs and may also trigger uploads to files.dlazy.com, so user data, secrets, or unrelated prior context could be unintentionally exfiltrated to external services.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.