Magnetic

• Outcome: Review
• Latest audit: May 28, 2026 at 10:20 AM UTC
• Version:
  1.0.4

Overview

This looks like a real Magnetic/Membrane integration, but it grants broad authenticated CRM/API control with unclear boundaries and some inconsistent service documentation.

Install only if you trust Membrane and intend to give the agent authenticated access to Magnetic data. Use the least-privileged Magnetic account available, verify the target domain before connecting, and require explicit approval before any create, update, delete, or raw proxy API request.

SkillSpector

By NVIDIA

Vulnerability Patterns

• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Description-Behavior Mismatch

Medium

• Confidence: 92% confidence
• Finding: The skill is presented as a narrowly scoped Magnetic integration, but the instructions enable broad connection management and arbitrary proxied API requests. That mismatch weakens least-privilege expectations and could let an agent perform actions outside the user’s understood intent or outside Magnetic-specific workflows.

Intent-Code Divergence

High

• Confidence: 97% confidence
• Finding: The skill claims to target Magnetic CRM functionality, but the described resource model and referenced domain appear inconsistent with that product and instead resemble an unrelated file/workspace system. This inconsistency is dangerous because it can mislead an agent into connecting to or operating on the wrong service, causing unauthorized access, data exposure, or unintended actions in a different application.

VirusTotal

64/64 vendors flagged this skill as clean.