Ab Tasty

• Outcome: Review
• Latest audit: May 29, 2026 at 9:36 PM UTC
• Version:
  1.0.2

Overview

This appears to be a legitimate A/B Tasty integration, but it gives agents broad authenticated access that could change or delete live experimentation data without clear confirmation guardrails.

Install only if you trust Membrane and need A/B Tasty automation. Use least-privilege A/B Tasty access, prefer discovered Membrane actions, and require explicit user approval before any POST, PUT, PATCH, or DELETE request.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• 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

Findings (2)

Vague Triggers

Medium

• Confidence: 87% confidence
• Finding: The skill description is broad enough that an orchestrator could invoke it for generic requests about 'manage data, records, and automate workflows,' even when the user did not specifically intend to operate on A/B Tasty. In a connected environment with write-capable actions, overbroad routing increases the chance of unintended data access or modification in the wrong system.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The proxy request section explicitly documents arbitrary direct API calls, including POST, PUT, PATCH, and DELETE, but does not warn that these methods can create, alter, or delete production experimentation data. That makes it easier for an agent to bypass safer prebuilt actions and perform destructive requests without confirmation or guardrails.

VirusTotal

63/63 vendors flagged this skill as clean.