ClawHub
Back to skill

Security audit

IMINI ImageGen

Security checks for vulnerabilities and agentic risk

Overview

This skill generates images through the imini API and saves the returned image files locally, and those behaviors are disclosed and aligned with its purpose.

Install only if you trust the imini image API and are comfortable sending prompts and public reference image URLs to it. Keep IMINI_IMAGE_API_KEY in local environment or skill config, verify the configured base URL, and expect generated files to be saved under output/imagegen unless you choose another output directory.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
91% confidence
Finding
The skill clearly uses environment variables, network access, and writes files, but it does not declare corresponding permissions in a dedicated permission model. That creates a transparency and governance gap: operators may approve or invoke the skill without understanding that it can exfiltrate data via network calls or persist files locally.

Tp4

High
Category
MCP Tool Poisoning
Confidence
84% confidence
Finding
The documented behavior promises a user choice of model before execution, but the finding indicates the implementation may silently choose a default model and also download/save remote results locally. This mismatch is dangerous because users and reviewers may authorize one workflow while the actual code performs additional actions or makes autonomous choices that affect cost, privacy, and local state.

Vague Triggers

Medium
Confidence
95% confidence
Finding
The skill enables implicit invocation (`allow_implicit_invocation: true`) without any activation constraints in the file itself, which increases the chance the agent will call an image-generation capability when the user did not explicitly request tool execution. Because this skill can send prompts and public image URLs to an external API, unintended invocation can cause unintended data disclosure, unexpected external actions, or policy bypass through over-broad tool selection.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The documentation explicitly expects the bundled script to download every returned image URL and write files locally, but provides no user-facing warning or consent checkpoint for network transfer and disk writes. In an agent context, this can cause surprising external fetches and local persistence of untrusted remote content, increasing risk of unsafe file handling, privacy issues, and unintended resource use.

Static analysis

No suspicious patterns detected.