ClawHub
Back to skill

Security audit

TinyPNG

Security checks for vulnerabilities and agentic risk

Overview

This TinyPNG skill does what it claims overall, but it under-labels image upload and transformation actions as safe read-like operations.

Install only if you are comfortable with agents using an OOMOL-connected TinyPNG account to upload or transform images. Treat `shrink_image` and `output_image` as mutating/external-processing actions, and confirm the exact image or URL before running them.

SkillSpector

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 (3)

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The manifest says the skill is for "searching and reading data," but the documented TinyPNG actions actually create or transform image resources. This misclassification can cause an agent or user to treat state-changing operations as harmless read-only actions, reducing scrutiny and confirmation before execution.

Intent-Code Divergence

High
Confidence
98% confidence
Finding
The safety section states that untagged actions are reads, but the only listed untagged actions (`output_image`, `shrink_image`) are processing operations that create or transform resources. This is dangerous because it directly instructs an agent to run mutating operations without confirmation under the false assumption that they are safe read-only calls.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The trigger language says to use this skill for "ANY TinyPNG request," which is overly broad and can cause unintended invocation whenever TinyPNG is mentioned, even if the user did not ask to operate on images. In the context of a skill capable of uploading and transforming image data, over-triggering increases the risk of unnecessary external data transfer and unintended actions.

Static analysis

No suspicious patterns detected.