Pirate Weather

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed Membrane-based Pirate Weather integration, with broad but visible API helper commands and no hidden code or destructive behavior found.

Install only if you are comfortable installing the Membrane CLI, logging into Membrane, and letting the agent make authenticated Pirate Weather API calls. Prefer discovered Membrane actions over raw proxy requests, and review any requested path, method, headers, and body before approving.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Description-Behavior Mismatch

Medium

Confidence: 91% confidence
Finding: The skill is scoped and described as a Pirate Weather integration, but the documented connection flow can create or discover connections generically by URL/domain and may auto-create an app/connector if none exists. That broadens the effective capability beyond a narrowly scoped weather skill, increasing the chance an agent could be steered into interacting with unintended services or capabilities.

Description-Behavior Mismatch

Medium

Confidence: 94% confidence
Finding: The proxy section exposes arbitrary HTTP methods, headers, body data, query parameters, and path parameters, which effectively enables broad raw API access rather than a narrowly bounded weather-data skill. Even if intended for flexibility, this expands the attack surface and can permit unexpected operations if the remote service or connector supports more than read-only forecast retrieval.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal