Blend

• Outcome: Review
• Latest audit: May 29, 2026 at 9:37 PM UTC
• Version:
  1.0.2

Overview

This Blend integration is coherent, but it gives an agent broad authenticated ability to change Blend data without clear confirmation guardrails.

Install only if you trust the Membrane CLI and are comfortable giving an agent delegated access to your Blend account. Prefer discovered Membrane actions, use the least-privileged account available, and require explicit approval before any create, update, publish, or delete operation.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (1)

Missing User Warnings

Medium

• Confidence: 89% confidence
• Finding: The skill documents a generic proxy request mechanism that supports mutating HTTP methods like POST, PUT, PATCH, and DELETE without any explicit caution about side effects, confirmation requirements, or read-only defaults. In an agent context, this increases the chance of unintended record changes or deletions against the user's Blend-connected data, especially when the model falls back to raw API calls.

VirusTotal

60/60 vendors flagged this skill as clean.