ClawHub
Back to skill

Security audit

Adobe Commerce

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed Adobe Commerce connector skill for reading product and category data through OOMOL, with some broad wording users should treat carefully.

Install only if you are comfortable giving the OOMOL connection read access to Adobe Commerce product and category data. Treat this version as read-only: use only get_category, get_product, list_categories, and list_products unless you separately review and approve any additional connector actions.

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
94% confidence
Finding
The manifest and top-level description position this skill as suitable for 'searching and reading data', but the body explicitly describes support for write and destructive actions. That mismatch can cause an agent or user to invoke the skill under a read-only trust assumption while still being capable of state-changing operations, increasing the risk of unintended modifications to Adobe Commerce data.

Intent-Code Divergence

Medium
Confidence
91% confidence
Finding
The instructions tell the agent to inspect schema and run an arbitrary action by name on the connector, while the safety section only informally states that untagged actions are reads. Because execution is generic and not technically constrained to the listed read actions, a caller could invoke undisclosed or future write-capable actions through the same connector path, bypassing the apparent safety framing.

Vague Triggers

Medium
Confidence
87% confidence
Finding
The trigger text says to use this skill for ANY Adobe Commerce request, which is broader than the documented read-oriented description and encourages over-routing. In practice, this can cause the agent to prefer this skill even for tasks outside its safe or intended scope, amplifying the effect of the capability mismatches elsewhere in the file.

Static analysis

No suspicious patterns detected.