Serpdog

• Outcome: Review
• Latest audit: May 29, 2026 at 9:38 PM UTC
• Version:
  1.0.4

Overview

This appears to be a legitimate Serpdog integration, but it gives an agent broad authenticated API access that could change or delete Serpdog account data without explicit guardrails.

Install only if you are comfortable connecting Serpdog through Membrane and letting the agent operate with that delegated access. Prefer discovered Membrane actions, and require the agent to show and confirm any raw proxy request before sending sensitive data or using POST, PUT, PATCH, or DELETE.

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: 87% confidence
• Finding: The skill encourages arbitrary proxy requests to an external API without explicitly warning that user-supplied data may be transmitted to Serpdog via Membrane. In an agent setting, this can lead to unintended disclosure of sensitive prompts, identifiers, or business data if the model constructs proxy calls from user content without clear confirmation or scoping.

VirusTotal

63/63 vendors flagged this skill as clean.