ClawHub
Back to skill

Security audit

Gift Up

Security checks for vulnerabilities and agentic risk

Overview

The skill is a legitimate Gift Up integration, but its safety labels understate several money-affecting gift card actions.

Install only if you are comfortable giving the agent access to Gift Up through OOMOL, and treat all gift card balance, redemption, void, reactivation, and undo actions as requiring explicit review of the payload and effect before execution. The publisher should correct the action labels before this is considered low-risk.

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)

Intent-Code Divergence

Medium
Confidence
98% confidence
Finding
The action inventory and safety model are internally inconsistent: `get_order` is incorrectly marked `[write]` while multiple clearly state-changing operations such as `reactivate_gift_card`, `redeem_gift_card`, `top_up_gift_card`, `undo_gift_card_redemption`, and `void_gift_card` are left untagged. An agent following the documented rule that untagged actions are safe reads could execute real money-affecting or destructive Gift Up operations without confirmation, causing unauthorized balance changes, redemptions, or card invalidation.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The instruction to use this skill for "ANY Gift Up request" is an overly broad routing trigger that can cause automatic invocation in situations where the user's intent is ambiguous or where direct clarification should happen first. In a commerce/gift-card context, broad auto-selection increases the chance that an agent reaches sensitive read/write capabilities before establishing whether the user wants information retrieval, a transaction, or a potentially destructive account action.

Missing User Warnings

Medium
Confidence
99% confidence
Finding
Several destructive or state-changing operations are presented without corresponding warnings despite the safety section asserting that untagged actions are safe reads. In this skill's context, untagged actions can redeem value, top up balances, undo transactions, or void cards, so the missing warnings materially increase the risk of unauthorized financial actions and business disruption.

Static analysis

No suspicious patterns detected.